Mineral Processing and Extractive Metallurgy Review

An International Journal

ISSN: 0882-7508 (Print) 1547-7401 (Online) Journal homepage: http://www.tandfonline.com/loi/gmpr20

True flotation versus entrainment in reverse

cationic flotation for the concentration of iron ore
at industrial scale

Virginia P. S. Nykänen, André S. Braga, Thiago C. Souza Pinto, Patricia H. L. S.

Matai, Neymayer P. Lima, Laurindo S. Leal Filho & Marisa B. M. Monte

To cite this article: Virginia P. S. Nykänen, André S. Braga, Thiago C. Souza Pinto, Patricia H. L.
S. Matai, Neymayer P. Lima, Laurindo S. Leal Filho & Marisa B. M. Monte (2018): True flotation
versus entrainment in reverse cationic flotation for the concentration of iron ore at industrial scale,
Mineral Processing and Extractive Metallurgy Review, DOI: 10.1080/08827508.2018.1514298

To link to this article: https://doi.org/10.1080/08827508.2018.1514298

Published online: 10 Oct 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=gmpr20
MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW
https://doi.org/10.1080/08827508.2018.1514298

True flotation versus entrainment in reverse cationic flotation for the concentration
of iron ore at industrial scale
Virginia P. S. Nykänena, André S. Braga a,b, Thiago C. Souza Pintoa, Patricia H. L. S. Mataib, Neymayer P. Limac,
Laurindo S. Leal Filhoa,b, and Marisa B. M. Montea,d
a
Instituto Tecnológico Vale, Minas Gerais, Brasil; bLaboratory of Transport Phenomena and Chemistry of Interfaces, Department of Mining and
Petroleum Engineering, University of São Paulo, São Paulo, Brazil; cVale S.A., Prédio 1, Mina de Águas Claras, Minas Gerais, Brasil; dLaboratory for
Surface Chemistry, Coordination of Mineral Processing – Center for Mineral Technology, CETEM, Avenida Pedro Calmon, 900, Cidade Universitária,
Rio de Janeiro, Brasil

ABSTRACT KEYWORDS
The loss of iron-bearing minerals to the tailings, especially in the finest fractions (−44 μm), is a problem Flotation; entrainment; true
that must be endured due to the depletion of deposits containing high-Fe hematite. The two main flotation; iron ore
mechanisms for iron mineral loss in the froth in reverse cationic flotation are hydrodynamic dragging
and true flotation. This last one being attributed to failures in the conditioning processes regarding the
depression of iron mineral by starch. In this work, two industrial mechanical flotation circuits located in
Minas Gerais—Brazil, namely Conceição Itabiritos II and Pico, operating with mineralogically distinct iron
minerals, had their tailings analyzed in order to attribute reasons for the iron-bearing minerals losses.
Liberation studies excluded losses due to the presence of composite particles where hematite was
combined to silica or goethite. The method known as two-liquid flotation was applied to the several
particle fractions present in the tailings to evaluate the particles surfaces hydrophobicity/hydrophilicity.
The results indicated that for Conceição Itabiritos II, the loss of iron minerals was due to a combination
of true flotation and hydrodynamic dragging. True flotation probably took place because of depression
failure. For Pico, the results revealed that most of the iron-bearing particles were lost due to hydro-
dynamic dragging. The shape factor and terminal velocity, both obtained via permeametry, confirmed
the two-liquid flotation results. Two-liquid flotation was shown to be a fast and simple method to
qualitatively assess the hydrophobicity of particles in froth flotation, thus allowing quick improvements
in the process.

1. Introduction depressant for iron ore is corn starch, due to the satisfactory
results delivered, and its great commercial abundance. The
During the first half of the 20th century iron was recovered
mechanism by which starch and hematite interact, in order to
primarily from high grade hematite bodies containing over 60
avoid the action of collector over the mineral surface, has
wt% Fe. To date, deposits containing high-Fe hematite have
been extensively discussed and can be checked elsewhere
been depleted and the mineral industry needs more often to
(Peres and Correa 1996; Pavlovic and Brandão 2003). The
deal with poorer quality hematite (banded iron formations—
collectors normally used are etheramines or etherdiamines,
BIFs) (Selmi et al. 2009; Gomes et al. 2015). BIFs are mechani-
which act also as frothers because of their carbon chain (Lima
cally more fragile than typical hematite deposits increasing the
et al. 2013).
amount of fines within the feed. Besides, to recover low-grade
The recovery of iron-bearing minerals particles in froth
minerals, those need to be ground until reaching a certain size
flotation is done either by true flotation or by entrainment
to ensure the mineral liberation.
(Eisele and Kawatra 2007; Lima et al. 2016). The true flotation
Low-grade iron ores are beneficiated in large scale by froth
refers to the process where particles are led to the froth layer
flotation in the same way as fine size range iron ores (<149
because of the hydrophobicity of their surface (Ross 1991).
μm) (Lima et al. 2013). Studies related to iron ore flotation
Commonly, the hydrophobicity of mineral particles is evalu-
began in 1931 and included the direct anionic flotation, and
ated through contact angle measurements, which involve
the cationic flotation of quartz, also called reverse cationic
proper equipment and sample preparation process, depending
flotation. In this process, the gangue mineral, mainly quartz,
on the method chosen (Chau 2009).
has its surface modified by a collector becoming hydrophobic,
Entrainment is a physical process where fine particles are
therefore it reported to the froth phase. At same time, iron
dragged to the froth layer, no matter whether their surface is
ore, under the action of a depressant agent, remains hydro-
hydrophilic or hydrophobic. These particles enter the lamella
philic at the bottom of the flotation cell. The most common

CONTACT Marisa B. M. Monte [email protected] Centro de Tecnologia Mineral, Avenida Pedro Calmon, 900, Cidade Universitária, RJ, Rio de Janeiro, Brasil.
21941-908
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gmpr.
© 2018 Taylor & Francis Group, LLC
2 V. P. S. NYKÄNEN ET AL.

2007). Therefore, this method could evaluate the effectiveness

of collector and depressant over a given mineral, during the
flotation process. Briefly, this method involves two solvents
with different polarities, i.e. polar and nonpolar, which are
mixed together with the particles to be separated, forming
two phases after ceasing the movement. Particles with hydro-
phobic surfaces will migrate to the nonpolar phase, attached to
the small bubbles of solvent, whereas particles with hydrophilic
surfaces will migrate and disperse into the polar phase
(Kocabağ and Güler 2007). If the particles present the same
surface properties for a particular system, surfactants or other
modifying agents can be used to tune the hydrophobic/hydro-
philic nature, so that the two-liquid flotation could be used also
Figure 1. Graphic representation of the entrainment phenomenon. Fine particles
for those cases. Two-liquid flotation provides fast results using
are dragged along and become entrapped within the lamellae in the froth. True very simple glassware and equipment.
floated particles are attached to the air bubble by the action of a collector. In this work, we analyzed the reasons for iron minerals being
recovered from the tailings froth in reverse cationic flotation of
iron ore. Two industrial flotation circuits, operating with miner-
of the froth becoming entrapped in this interbubble water alogically distinct iron ores were object of our studies, using
layer (Figure 1). The entrainment dilutes the desired material, mainly the two-liquid flotation and permeametry to evaluate in
reducing the grade of the product (Neethling and Cilliers a simple, but robust way the source of valuable minerals loss to the
2009; Wang et al. 2015). The use of a proper froth washing tailings. Both sites, Mina do Pico and Mina Conceição Itabiritos II,
system in flotation cells can decrease the contribution of and their flotation circuits, are located at Minas Gerais, Brazil, in
entrainment in the loss of fine particles; however, those sys- the region known as “Quadrilátero Ferrífero,” and they are part of
tems are more often a feature of column flotation cells two distinct mineral complexes of iron ore exploitation belonging
(Kawatra and Eisele 1987; Dowling et al. 2000; Shean and to the Brazilian mining company Vale S/A.
Cilliers 2011).
It is well known that fine particles usually contain the
highest amount of iron-bearing minerals lost to the tailings 2. Materials and methods
(Gomes et al. 2011). Although fine particles can also take part 2.1. Sample collection and preparation
in true flotation, those tiny particles are especially affected by
the entrainment phenomenon, mainly due to their lack of Samples were collected in duplicates on the automatic cutter
mass. In this context, finding out whether true flotation or samplers from the industrial flotation circuits, during a survey
entrainment is the predominant reason for valuable mineral of 2 h, with increments being collected every 10 min to
loss could improve the operation parameters in order to compose the final sample. The sampled points on the circuit
minimize the loss (Schubert 2008). were: fresh feed, final concentrate, and final tailings. Both
The entrainment of particles within the froth layer will circuits operated with conventional flotation machines, with
depend on many factors; among them are the size and shape no froth washing system. Samples were sized into three frac-
of those particles (Wiese et al. 2015). The parameter asso- tions (+105 μm, −105 + 44 μm, −44 μm) for the two-liquid
ciated to the shape of particles is the sphericity or shape factor flotation and permeametry experiments. After those proce-
(ψ), which gives information about how rounded is a deter- dures, samples from the feed, concentrate, and tailings were
mined particle (Mauri 2015). The highest value possible for sent to chemical and morphological characterization.
this parameter, (Ψ = 1), corresponds to a perfect sphere
(Mauri 2015; Souza Pinto et al. 2016).
2.2. Mineral characterizations
Permeametry is a technique that allows us to assess the
sphericity of a particle size distribution (Schulz 1974). 2.2.1. XRF (X-ray fluorescence)
Permeability is a characteristic parameter of a porous media, X-ray fluorescence was used to determine the chemical com-
being related to the arrangement and distribution of pores by position of the samples. All the samples were analyzed as
which a given Newtonian fluid must pass through. In the melted pellets, and the loss to fire was analyzed by gravimetry,
permeametry, a porous media (bed of particles near the for 1 h at 1050°C.
same size) is introduced in a close-fitting tube in order to
avoid fluid leakage in between the bed of particles and the 2.2.2. Mineralogical, chemical, and morphological analyzes
walls of the tube. This population of particles can have its Mineralogical analyzes were done by X-ray diffraction (XRD),
sphericity determined by the percolation of a gas passing using the powder method and a position sensitive detector.
through it, in a laminar flux (Schulz 1974; Eriksson et al. Crystalline phases were compared with the following data-
1990; Souza Pinto et al. 2009). bases: ICDD (International Center for Diffraction Data)—
Two-liquid flotation is a notably useful method to separate 2003 and PANICSD (PANanalytical Inorganic Crystals
fine particles, normally smaller than 10 µm, according to their Structure Database)—2007. Qualitative and semiquantitative
hydrophobicity (Kocabağ and Güler 2007; Otsuki and Dodbiba evaluation of chemical elements within the samples was
 MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW 3

conducted by scanning electron microscopy (SEM)— iron contained in the flotation feed (57%) is concentrated in the
Stereoscan 400 (LEO)—equipped with an energy dispersion particles < 30 μm. The iron particles higher than 150 μm
scattering detector (EDS)—Oxford. The analyzed sections represent only 3.9% of flotation feed and 1.1% of these particles
were previously sputtered with carbon. Samples having dia- were reported to the froth (tailings). On the other hand, the
meters over 20 μm were subject of detailed mineralogical percentage of SiO2 (55%) particles more expressive was found in
study to determine the mineral occurrence, and the associa- the coarse size fractions (+150 μm) (Figure 2b). It was also
tions of those minerals. The analyses were done by the use of observed that the SiO2 particles recovery having −150 + 75 µm
the software MLA—Mineral Liberation Analyzer—(FEI), over and −75 + 30 µm were very high (~96% and 99%, respectively).
polished surfaces of the particles. The software was coupled to Figures 3a and 3b shows the Fe and SiO2 distributions in
a SEM—600 (FEG)—which has also a microanalyses system feed, concentrate, and tailing, respectively, for Pico plant, as a
EDS Espirit—Brucker. function of particle size. It was observed that 12% of iron
particles smaller than 30 µm floated. Pure hematite fine par-
2.2.3. Hydrophobicity test—two-liquid flotation ticles are mostly unfloatable (25%), as expected, and this
For the two-liquid flotation experiments, 375 mL of distilled significant percentage of particles (12%) that float is probably
water and 125 mL of isooctane were poured into a 1000 mL there due to hydrodynamic drag and entrainment processes
beaker, and the mixture was mechanically stirred for 2 min at and/or lack of depressant. It can be seen from Figure 3b that
1000 ± 20 rpm, to condition the mixture. The rotor size SiO2 recovery in the concentrate was very low and the tailing
choice followed the relation Dbeaker = 3 × drotor, which gave contains mainly coarse SiO2 particles.
d = 37.08 mm (Froude number = 1.05). These experimental Although we have conducted studies for samples acquired
conditions adopted provided a very well homogenous mixture from feed, concentrate, and tailings, our major interest was in
without vortex creation (Arbiter et al. 1976). After the con- regard to the contents of iron-bearing minerals in the tailings,
ditioning time, ~ 20 g of a sample taken from the fraction −44 besides the size and shape of those minerals particles. Here, the
µm (final tailings), were added to the stirring mixture and iron minerals dealt with are basically hematite (Fe2O3) and
stirred for 3 min, with that same rotor speed. It is important magnetite (Fe3O4), and less importantly, the hydrated iron
to point out that the mineral samples must still be humid; not oxide goethite (FeO)OH. From now, to facilitate the reading,
passing by any drying processes prior the experiment, to keep we shall assign the results as FTC (flotation tailings from
the chemical environment in which the particles were Conceição Itabiritos II) and FTP (flotation tailing from Pico).
immersed in the industrial cell/tank. The stirring was stopped, Silica, mostly present as quartz mineral is typically the most
and the system was kept untouched for 3 min to allow proper abundant contaminant for the iron ores, despite other minerals,
migration of the particles to the preferred chemical environ- which are present in much lower concentration. Figures 4a and
ment. After the clear separation of the two phases, the upper 4b brings the iron and silica contents in the particle size frac-
one was carefully collected with help of a Pasteur pipette to tions analyzed for both FTC and FTP. Not surprisingly, an
avoid loss of the phase. Both, collected and remaining mate- inspection of the graphs reveals that the largest loss of iron, in
rial were filtered, dried in an oven at 100°C, and weighed to terms of iron contents for both cases, was found in the finest
obtain the floated mass and the sunk mass. This test was fractions. The particles smaller than 44 μm comprise about 71%
conducted in quadruplicate. of all the iron lost in the flotation process of Conceição Itabiritos
II plant and 85% for the Pico plant.
2.2.4. Sphericity test—permeametry Considering all particle sizes, i.e. all the particles larger
For permeametry experiments, three size fractions (+105 μm, than 20 μm, it was found that FTP was composed basically
−105 + 44 μm, −44 μm) were magnetically separated to render of hematite (16%), quartz (80%), and goethite (3.7%). FTC
two samples: magnetic and nonmagnetic. After that, each was composed of hematite (14%), quartz (85%) and, in this
sample (magnetic and nonmagnetic from final tailings) was case, the goethite represented only 0.7% of the minerals pre-
packed in a small Büchner funnel, measuring 48.5mm in sent in the tailings (Table 1). The larger amount of goethite in
diameter and 23.3 mm in height. The packed funnel bed FTP shows the mineralogical difference between the iron ores
was inserted in a Kitassato glass, which had two lateral con- extracted from these distinct mines.
nections: (i) one was linked to a flowmeter coupled to a Goethite is a very common mineral, and it can present
vacuum pump (200 mmHg); (ii) another went to a man- several shapes and sizes, occurring mostly as a needle-like
ometer for pressure loss measurements. crystal. The presence of goethite makes the ore more friable
The permeametry tests were conducted by varying the increasing the amount of fines after grinding. Among the
suction flow from 0 to 5 L.min−1. The relationship between types of goethite, the earthy-like one is characterized by
pressure drop and flow allows the calculation of the permea- higher amounts of Al2O3 and SiO2, together with other impu-
metry constant (K), which is the parameter for calculating the rities (Santos and Brandão 2003). Differently, FTC was poorer
sphericity (Ψ). than FTP in goethite contents rendering larger particles after
the process as will be shown later in this work.
3. Results and discussion
3.1. Liberation studies
Figures 2a and 2b shows the Fe and SiO2 distributions in feed,
concentrate, and tailing, respectively, of Conceição Itabiritos II Based on the already presented results, we concentrated our
plant, as a function of particle size. The higher percentage of studies on the finest fractions, from where most of the iron
4 V. P. S. NYKÄNEN ET AL.

(A) 56.7
150 µm
60 48.0
75 µm

Fe Distribution (wt%)
50 39.4 30 µm
37.0
40

30
8.6
20
3.9 2.4 30 µm
10
2.8
1.2 75 µm
0
Flotation Feed
150 µm
Concentrate
Tailings

(B) 150 µm
60 54.5
53.1 75 µm
SiO2 Distribution (%)

50 37.1
36.6 30 µm
40
8.4
30
0.4 8.1
20
0.4 30 µm
10
1.4 75 µm
0
Flotation 150 µm
Feed Concentrate
Tailings

Figure 2. Metallurgical data as a function of particle size of Conceição Itabirito II plant. (a) Fe recovery (wt%) and (b) SiO2 recovery (wt%).

minerals had been lost to the froth. There are some hypoth- Table 4 presents the liberation class of hematite, and its
eses, which should be contemplated in order to attribute the associations, with respect to the particle size fraction for FTC.
right source of loss, so the process can be accordingly tuned to In this case, the particles having −44 + 20 μm presented
minimize those losses. complete liberation (96 wt%), since particles having 95 wt%
The first possibility for the iron loss that must be taken liberation are considered liberated. The class of liberation
into account is its association to other minerals forming decreases with the increase in the size of the particles, as
composites that could have two or more mineral phases. If expected. For the complete range of particle sizes, +20 μm,
the mineral associated to the iron ore is susceptible to the the liberation was 83 wt%.
action of the collector, there will be losses due to true flota- The two-phase composites represented 17 wt% of the
tion, according to the size of the phase containing the con- hematite with the second phase composed basically of quartz
taminant mineral. In the case of iron ore, the composite (13 wt%) or goethite (3.0 wt%). The three-phase composite
particles are mainly composed of quartz, besides hematite of was composed most often of hematite, quartz, and goethite.
course. Other minerals are always present, but in negligible This last one representing only 0.4 wt%. These results raise
quantities to interfere in the flotation process. the hypothesis of a depressant failure instead of collector’s
SEM micrographs associated with the MLA software data action over the quartz present on the composites.
were used to evaluate the degree of liberation of the iron Table 5 brings the results of the liberation studies for
carrier particles. Figure 5 shows representative features of hematite, and its associations, with respect to the particle
the images acquired from samples of FTC and FTP. size fraction for FTP. According to these results, the finest
Figure 5a (FTC) shows the image of a quartz particle having size range particles were not liberated as observed for FTC.
hematite associated as an inlaid particle. At Figure 5b (FTP), However, for the complete range of particles, the liberation
the typical features of goethite can be noticed, being more was basically the same as found for FTC, 82 wt%. The binary
fibrous and porous than hematite or quartz particles. Tables 2 composites represented 17 wt% of the hematite, being the
and 3 present the chemical composition, revealed by EDS, of second phase composed of goethite (13 wt%) and
the spots assigned with numbers in Figures 5a and 5b. quartz (3.4%).
 MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW 5

49 150 µm
(A) 50
47
37
75 µm

Fe Recovery (wt%)
40
25 30 µm
30
12
20 14
13
10 2 30 µm
1 75 µm
0
150 µm
Flotation Feed
Concentrate
Tailings

(B) 150 µm
60 57
55 75 µm
SiO2 Recovery (wt%)

50 30 µm
38
37
40

30

5
20
1 4
10 1 30 µm
2
75 µm
0
150 µm
Flotation Feed
Concentrate
Tailings

Figure 3. Metallurgical data as a function of particle size of Pico plant. (a) Fe recovery (wt%) and (b) SiO2 recovery (wt%).

The finest FTP particles did not present the same liberation iron-bearing minerals losses could be attributed to the true
as FTC particles, being considered not liberated. Despite this, flotation by means of the composite particles, where hematite
important information must be taken into account before is associated to quartz. Besides, larger particles are less
assuming that true flotation has taken place due to lack of affected by hydrodynamic dragging.
liberation. FTP binary composites were composed mainly of
associations between hematite and goethite, which is also an
iron-bearing mineral. Therefore, the goethite phase very likely
3.2. Permeametry—size and shape of particles
did not interfere in the depression of the iron minerals
because, at least in theory, the collector should exhibit basi- Beyond the size of the particles present in a flotation tank cell,
cally the same behavior toward hematite and goethite. their shape is also a crucial factor. Depending on its shape, a
Moreover, possible effects arising from interactions between particle can markedly experience the entrainment mechanism
collector and such composites would be diminished for the due to the ascendant flux in the system, and to the energy
amount of quartz is much lower than goethite. released by the impeller within the flotation cell (Smith and
For the finest particles, the results of the liberation studies Warren 1989). Thus, besides the size, it is essential to introduce
for both, FTC and FTP, indicate a depressant failure in addi- the shape factor, given by the sphericity function of the particles,
tion to hydrodynamic dragging for the loss of iron to the in the investigations about the loss of iron bearing minerals to
froth, instead of an interference of a silica collector provoking the flotation froth. The sphericity can assume values between
true flotation. On the contrary, for the coarser class size, zero and one (0 < Ψ ≤ 1), and as it gets closer to unity it gets
which presented a degree of liberation around 10 wt%, the more similar to a perfect sphere (Wadell 1935). A sphere is easier
6 V. P. S. NYKÄNEN ET AL.

Figure 4. Contents of Fe and SiO2 in the particles analyzed from (a) FTC and (b) FTP, according to their size distribution.

Table 1. Mineral composition of the tailings from the flotation process of to drag than a tetrahedron, for instance. Figure 6 illustrates some
Conceição Itabiritos II (FTC) and Pico (FTP) plants.
common shapes and the sphericity associated to them.
FTC FTP Considering very low Reynolds numbers, the pressure drop
Total (+20 μm) Total (+20 μm) for the laminar air flow can be calculated by using the
Minerals (%) (%)
Kozeny-Carman relation (Equation 1) (McCabe et al. 2004):
Hematite 14 16
Quartz 85 80
;
Goethite 0.7 3.7
ΔP 72λ1 V o μð1  εÞ2
Clay minerals 0.2 0.2 ¼ (1)
Others 0.1 0.1 L Ψ2 D23;2 ε3
 MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW 7

Where, ΔP/L = press ure drop through a packed bed of length

L; λ1 = correction factor equal to 2.5, or 72 λ1 = 180 (Mauri
2015); v0 0 =average fluid velocity; µ = dynamic air viscosity; ε
= porosity; Ψ = sphericity of particles; and D3,2 = Sauter mean
diameter of particles.
For a size distribution of particles, the permeability of a
medium (K) can be determined by Equation 2, which is a
rearrangement of Darcy’s law (McCabe et al. 2004):
μQ
K¼ (2)
AΔP=L
Where, K = permeametry constant; µ = dynamic air viscosity; Q =
airflow rate; A = cross-sectional area of packed bed; ΔP/L =
pressure drop through a packed bed of length L.
Using Equations 1 and 2, it is possible to calculate the
sphericity (Ψ) according to Equation 3 (McCabe et al. 2004;
Mauri 2015):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
180K ð1  εÞ2
Ψ¼ (3)
D23;2 ε3

Where, K = permeability of porous media, D3,2 = Sauter mean

diameter of particles, and ε = porosity of the column of
particles. Tables 7 and 8 show the D3,2 and sphericity results
of hematite and quartz particles for FTC and FTP, respec-
tively. Note that Sauter’s diameter was determined by laser
diffraction in another experiment.
In Tables 6 and 7, it is possible to notice that the sphericity
increases more sharply for the finest particles (−44 µm), and it
keeps almost constant for the two other size fractions, not
showing a direct relation between D3,2 and Ψ for these range
of sizes. Unfortunately, it was impossible to determine the
sphericity of the quartz particles for −44 μm of FTP samples
because the pressure drop through a packed bed was too high
for the experimental apparatus used.
The terminal velocity (TV) of the hematite particles was
Figure 5. SEM micrographs acquired from representative samples of (a) FTC and (b) calculated following the method proposed by Concha and
FTP. The regions assigned with numbers were chemically analyzed by EDS and their
composition in terms of Fe, SiO2, Al2O3, and P is shown in Tables 2 and 3.
Barrientos (Concha and Barrientos, 1986) for nonspherical
particles, and it takes into account the shape factor (spheri-
city) of the particles. Nonspherical particles have different
behavior from spherical particles during sedimentation.
Table 2. Chemical composition of the regions assigned in the SEM image Whereas spherical particles follow basically a vertical trajec-
acquired from FTC sample, shown at Figure 3a. tory, the nonspherical ones rotate, vibrate, and describe a
Element/mineral grade spiral path during their fall. Still according to Concha and
(wt%) Barrientos, the sphericity by itself is not enough to determine
EDS region Fe SiO2 the shape of particles because there can be particles with same
1 – 100 sphericity, but different shapes. However, they have proved
2 68.7 –
that for isometric particles the sphericity can definitely dis-
criminate their shapes.
For particles which have the same density and diameter, it
Table 3. Chemical composition of the regions assigned in the SEM image
acquired from FTP sample, shown at Figure 3b.
is not a surprise that the terminal velocity is larger for parti-
Element/mineral grade
cles that are more spherical. For Pico Mine, TV of tailings
(wt%) particles is higher than those from concentrate because parti-
EDS region Fe SiO2 Al2O3 P cles of quartz (the main component in these tailings) have an
1 57.7 1.45 1.97 – intermediate shapes, i.e. neither needle nor spherical and it
2 57.3 1.44 2.34 0.11 seems that floated much better.
3 57.8 1.54 2.00 –
4 63.4 0.52 – – The particle shape has a significant effect on the floatability
5 63.0 0.71 0.44 – of mineral and hence the detachment force of shaped particle
6 – 100 – – is higher than that spherical particle (Xia 2017).
8 V. P. S. NYKÄNEN ET AL.

Table 4. Class of liberation and association forms of hematite, according to the particle size distribution, for FTC.
Particle size distribution
(μm)
Hematite associations
(wt%) Total (+20) +210 −210 + 150 −150 + 105 −105 + 74 −74 + 44 −44 + 20
Liberated 83 11 18 33 54 80 96
Composite (2 phases) 17 88 81 65 44 19 4.0
Composite (3 phases) 0.4 0.7 1.4 1.3 1.2 0.8 0.1

Table 5. Class of liberation and association forms of hematite, according to the particle size distribution, for FTP.
Particle size distribution
Hematite associations (µm)
(wt%) Total (+20) +150 −150 + 105 −105 + 74 −74 + 44 −44 + 20
Liberated 82 8.3 19 53 76 87
Composite (2 phases) 17 67 67 40 21 13
Composite (3 phases) 1.2 25 14 6.9 2.5 0.1

Figure 6. Relation between particle shapes and sphericity (ψ) values.

Table 6. Measured Sauter diameter (D3,2) and calculated sphericity (ψ) of hema- Figure 7 compares the terminal velocities found for the
tite particles, according to the size distribution for the tailings and concentrate
from Conceição Itabiritos II plant.
hematite particles reporting to the tailings compared to the
particles present in the concentrate, for three size fractions.
+105 μm −105 + 44 μm −44 μm
For both FTC and FTP we observed a reduction in TV
Flotation product Sample source D3,2 ψ D3,2 ψ D3,2 ψ
with the decrease in the size of the particles, which was
Tailings Pico (FTP) 120.1 0.50 73.4 0.47 12.2 0.77
Conceição (FTC) 55.8 0.56 53.4 0.52 16.2 0.72 expected. The results obtained for the same size range of
Concentrate Pico 103.4 0.37 59.6 0.48 36.1 0.50 particles collected in different points of the flotation cir-
Conceição 157.3 0.32 65.8 0.51 29.8 0.57
cuit, showed markedly different values for TV, which

Table 7. Measured Sauter diameter (D3,2) and calculated sphericity (ψ) of quartz particles, according to the size distribution for the tailings and concentrate from Pico
flotation plant.
+105 μm −105 + 44 μm −44 μm
Flotation product Sample source D3,2 ψ D3,2 ψ D3,2 ψ
Tailings Pico (FTP) 146.7 0.52 88.2 0.52 7.3 ND(*)
Conceição (FTC) 166.8 0.58 76.5 0.57 13.3 0.95
Concentrate Pico 276.2 0.31 ND(**) ND(**) ND(**) ND(**)
Conceição 309.0 0.53 91.1 0.46 20.8 ND(**)
(*) Not determined: with pecked bed highly compact, the apparatus was not able to measure the pressure drop.
(**) Not determined: the magnetic separation was not efficient.

strongly support that the particles reporting to the tailings

Table 8. Fe and SiO2 contents and distribution obtained from the finest particles
(−44 μm) at the hydrophobic and hydrophilic phase in the two-liquid flotation present different behavior than the particles in the con-
experiments conducted for FTP. centrate. This is especially true for FTC and corroborates
Final tail (−44 for the validation of the hypothesis that the loss of hema-
µm) Hydrophilic phase Hydrophobic phase tite particles happens through hydrodynamic dragging. In
Contents Dist. Contents Dist. Contents Dist. FTC, we noticed that the difference between TV calculated
Analyte (wt%) (%) (wt%) (%) (wt%) (%)
for tailings and concentrate particles, for the fraction +105
Fe 57.6 ± 0.3 100.0 57.6 ± 0.3 97.7 ± 1.2 56.3 ± 1.0 2.3 ± 1.5
SiO2 12.6 ± 0.2 100.0 12.7 ± 1.2 98.7 ± 1.0 6.5 ± 2.1 1.3 ± 1.0 μm, was very pronounced because of the difference in
 MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW 9

Figure 7. Terminal velocity (TV) calculated for the particles of hematite in the concentrate and final tailings of (a) Conceição Itabiritos II flotation plant and (b) Pico
flotation plant, according to the size distribution.

particle size (concentrate = 55.8 µm; tailings = 157.3 µm, froth flotation were effective or not. By chemical processes we
see Table 6), which decreases in finest (−105 µm) frac- mean, for instance, the depression of iron-bearing particles by
tions. The same was not true for FTP where the hematite starch, or even the collection of quartz particles.
particles found in the tailings presented higher TV values Two-liquid flotation, as already explained in the introduction
for the two larger size fractions, becoming lower than the of this work, is a separation method that gives quick answers
concentrate particles only for the finest particles. Those about the hydrophobicity/hydrophilicity of small particles. By
results are associated with the type of mineral and its weighing the contents of the two phases resulting from the experi-
associations, showing an intimate relation between term- ment, one can make qualitative conclusions about the surface
inal velocity and the mineralogical nature of the ore. properties of the minerals. The chemical analysis of each phase
Although the particles were the same size, they could enhances the quality of the results from qualitative to semiquanti-
contain more than one mineral phase (goethite) resulting tative. Thus, improving the reliability of the results.
in different density and TV values and to pertain for a Tables 8 and 9 present the results acquired from the two-liquid
same evaluated specific distribution of size. flotation experiments conducted for the particles of the size range

3.3. Hydrophobicity of the particles Table 9. Fe and SiO2 contents and distribution obtained from the finest particles
(−44 μm) at the hydrophobic and hydrophilic phase in the two-liquid flotation
According to the results acquired in the permeametry experi- experiments conducted for FTC.
ments, the fine particles could have reached the froth via hydro- Final tail (−44 µm) Hydrophilic phase Hydrophobic phase
dynamic dragging, becoming entrapped in the lamellae, as Contents Dist. Contents Dist. Contents Dist.
explained for the entrainment phenomenon. However, it is neces- Analyte (wt%) (%) (wt%) (%) (wt%) (%)
sary to evaluate the chemical nature of the particles’ surface to Fe 47.9 ± 0.9 100.0 43.0 ± 1.4 34.6 ± 2.4 51.0 ± 0.8 65.4 ± 2.4
SiO2 28.2 ± 1.0 100.0 34.9 + 1.6 47.6 ± 2.2 24.1 ± 0.8 52.1 ± 2.2
ensure if some failure during the chemical processes within the
10 V. P. S. NYKÄNEN ET AL.

−44 μm from FTC and FTP, respectively. As the experiment is besides the hydrodynamic dragging that seems to take
completed, the upper phase is composed of the isooctane and the place in both cases.
hydrophobized particles, and the lower phase, together with the
damped material corresponds to the hydrophilic particles
(Figure 8).
In principle, the iron present in the tailings should be 4. Conclusion
hydrophilic thus it should be collected from the lower In this work, iron ores coming from two different mines
phase. The hydrophilicity of iron-bearing particles confirms and their flotation plants were the subject of study in order
the effectiveness of the depression process by starch. The to evaluate if the loss of iron-bearing minerals to the tail-
iron-containing particles reporting to the upper layer ings happened predominantly due to hydrodynamic drag-
reached the tailings (froth phase) either by hydrodynamic ging or true flotation, in the reverse cationic flotation
dragging or depressant failure. In this last case, the cause process. The results obtained from the separation method
was an insufficient addition or inefficiency of the depres- called two-liquid flotation together with permeametry and
sion agent. liberation studies, for the finest particles (−44 μm), allowed
Beginning with FTP, according to Table 8, the recovery us to conclude that the loss of iron-bearing minerals
of iron from the upper hydrophobized phase was 2.30% occurred by different means for the two flotation circuits
whereas 97.7% of the Fe remained in the hydrophilic analyzed. For the FTC, flotation tailings from Conceição
phase. Regarding quartz, 1.30% was recovered from the Itabiritos II, the finest particles were considered liberated;
hydrophobic upper phase, and 98.7% of this mineral therefore, the loss did not occur due to the presence of
remained in the lower phase being considered hydrophilic. composite particles. In this case, the loss happened through
Combining these results with the permeametry results, it true flotation in addition to hydrodynamic dragging. True
can be suggested that, the finest particles of hematite flotation probably occurred due to a failure in the depres-
reached the froth phase by hydrodynamic dragging. Fine sion process of hematite with starch. For FTP, flotation
particles of quartz, in the same way, reported to the tail- tailings from Pico, the finest particles did not present the
ings also via hydrodynamic dragging, once they were not same degree of liberation shown by FTC, being considered
hydrophobic, as proven by their scarce presence in the not liberated. The binary composite particles, in this case,
hydrophobic phase. were mainly composed of hematite and goethite. As
The analysis of the data collected from FTC samples goethite should display the same behavior of hematite
showed a different behavior. As shown in Table 9, the toward the depressant (starch) and the quartz collector,
hydrophobic upper phase presented an iron recovery of the loss of iron-bearing minerals to the tailings probably
65.4%, whereas hydrophilic lower phase presented Fe recov- did not occur due the presence of such composites. For the
ery of 34.6%. This means that there were hydrophobic FTP, two-liquid flotation showed that the majority of iron-
hematite particles, thus they were not successfully bearing particles lost to the froth phase got there via hydro-
depressed. For quartz, the hydrophobic particles corre- dynamic dragging, once nearly all the iron-bearing particles
sponded to 52.1% and hydrophilic particles to 47.6%. remained in the hydrophilic phase. For the FTC, 65.4% of
Those results show that, for FTC, fine particles of hematite hematite is recovered in the hydrophobic phase demon-
could have reached the froth not only by hydrophobic strating that the loss of iron minerals was predominantly
dragging but also by true flotation. Moreover, a larger due to a true flotation. True flotation probably took place
amount of hematite particles underwent true flotation because of depression failure.
than quartz particles. Differently from FTP, where the In both cases, two-liquid flotation was crucial to assess the
principal way of losing iron-bearing minerals to the froth results, providing fast and reproducible results and operating
was found to be the hydrodynamic dragging, the hematite with simple glassware and equipment. This method could be
present in FTC reached the froth by true flotation means, easily applied in concentrator premises to provide quick eva-
probably due to inefficiency in the depression process, luations for the iron ore depression efficiency.

Figure 8. Typical appearance of the two phases during the resting time before the collection of the phases’ contents of (a) FTC and (b) FTP. Note the difference in the
color of the ores, especially in the upper phase, where FTP presents the typical color for goethitic minerals.
 MINERAL PROCESSING AND EXTRACTIVE METALLURGY REVIEW 11

Acknowledgment Lima, N. P., Valadão, G. E. S., and Peres, A. E. C., 2013, “Effect of amine
and starch dosages on the reverse cationic flotation of an iron ore.”
The authors acknowledge Vale S/A for the samples and technical Minerals Engineering, 45. pp. 180–184.
support. Mauri, R., 2015, Transport Phenomena in Multiphase Flows in Fluid
Mechanics and Its Applications, Vol. 112 vol., Switzerland: Springer
Ed., pp. 459.
Disclosure statement McCabe, W. L., Smith, J. C., and Harriott, P., 2004, Units Operation of
Chemical Engineering, 7th edition, New York: McGraw-Hill Ed., pp. 1140.
The authors report no conflicts of interest. The authors alone are
Neethling, S. J., and Cilliers, J. J., 2009, “The entrainment factor in froth
responsible for the content and writing of the article.
flotation: model for particle size and other operating parameter
effects.” International Journal of Mineral Processing, 93. pp. 141–148.
Otsuki, A., and Dodbiba, G., 2007, “Heterocoagulation of fine particles in
ORCID polar organic solvent.” Materials Transactions, 48(5). pp. 1095–1104.
André S. Braga http://orcid.org/0000-0002-2558-276X Pavlovic, S., and Brandão, P. R. G., 2003, “Adsorption of starch, amylose,
amylopectin and glucose monomer and their effect on the flotation of
hematite and quartz.” Minerals Engineering, 16(11). pp. 1117–1122.
References Peres, A. E. C., and Correa, M. I., 1996, “Depression of iron oxides with
corn starches.” Minerals Engineering, 9(12). pp. 1227–1234.
Arbiter, N., Harris, C. C., and Yap., R. F., 1976, “The air flow number in Ross, V. E., 1991, “The behaviour of particles in flotation froths.”
flotation machine scale-up.” International Journal of Mineral Minerals Engineering, 4(7). pp. 959–974.
Processing, 3. pp. 257–280. Santos, L. D., and Brandão, P. R. G., 2003, “Morphological varieties of
Chau, T. T., 2009, “A review of techniques for measurement of contact goethite in iron ores from Minas Gerais, Brazil.” Minerals Engineering,
angles and their applicability on mineral surfaces.” Minerals 16(11). pp. 1285–1289.
Engineering, 22(3). pp. 213–219. Schubert, H., 2008, “On the optimization of hydrodynamics in fine
Concha, F., and Barrientos, A., 1986, “Settling velocities of particulate particle flotation.” .Minerals Engineering, 21(12–14). pp. 930–936.
systems 4. Settling of nonspherical isometric particles.” International Schulz, N. F., 1974, “Measurement of surface areas by permeametry.”
Journal of Mineral Processing, 18(3). pp. 297–308. International Journal of Mineral Processing, 1(1). pp. 65–79.
Dowling, E. C., Hebbard, J., Eisele, T. C., and Kawatra, S. K., 2000, Selmi, M., Lagoeiro, L. E., and Endo, I., 2009, “Geochemistry of hema-
Processing of Iron Ore by Reverse Column Flotation, In Proceedings titite and itabirite, Quadrilátero Ferrífero, Brazil.” Rem: Revista Escola
of the XXI International Mineral Processing Congress, (P. Massaci, de Minas, 62(1). pp. 35–43.
Ed.), Amsterdam: Elsevier Ed., pp. 163–170. Shean, B. J., and Cilliers, J. J., 2011, “A review of froth flotation control.”
Eisele, T. C., and Kawatra, S. K., 2007, “Reverse Column Flotation of Iron International Journal of Mineral Processing, 100. pp. 57–71.
Ore.” Minerals and Metallurgical Processing, 24(2). pp. 61–66. Smith, P. J., and Warren, L. J., 1989, Entrainment of particles into
Eriksson, M., Nyström, C., and Alderborn, G., 1990, “Evaluation of a flotation froth, In Frothing in Flotation, (J. S. Laskowski, Ed.), New
permeametry technique for surface area measurements of coarse par- York: Gordon and Breach Science Publishers, pp. 123–144.
ticulate materials.” International Journal of Pharmaceutics, 63(3). pp. Souza Pinto, T. C., Lima, O. A., and Leal Filho, L. S., 2009, “Sphericity of
189–199. apatite particles determined by gas permeability through packed
Gomes, M. A., Pereira, C. A., and Peres, A. E. C., 2011, “Caracterização beds.” Minerals & Metallurgical Processing, 26. pp. 105–108.
tecnológica de rejeito de minério de ferro.” Rem: Revista Escola de Souza Pinto, T. C., Slatter, P. T., Matai, P. H., and Leal Filho, L. S., 2016,
Minas, 64(2). pp. 233–236. “The influence of hematite particle shape on stratification in pipe
Gomes, R. B., Tomi, G. D., and Assis, P. S., 2015, “Impact of quality of flow.” Powder Technology, 302. pp. 75–80.
iron ore lumps on sustainability of mining operations in the Wadell, H., 1935, “Volume, shape and roundness of quartz particles.”
Quadrilatero Ferrifero Area.” Minerals Engineering, 70. pp. 201–206. Journal of Geology, 43(3). pp. 250–280.
Kawatra, S. K., and Eisele, T. C., 1987, Column Flotation of Coal, In Fine Wang, L., Peng, Y., Runge, K., and Bradshaw, D., 2015, “A review of
Coal Processing, (S. K. Mishra and R. R. Klimpel, Eds.), Park Ridge: entrainment: mechanisms, contributing factors and modeling in flota-
Noyes Publications, pp. 414–429. tion.” Minerals Engineering, 70. pp. 77–91.
Kocabağ, D., and Güler, T., 2007, “Two-liquid flotation of sulphides: Wiese, J., Becker, M., Yorath, G., and O’Connor, C., 2015, “An investiga-
an electrochemical approach.” Minerals Engineering, 20. pp. 1246– tion into the relationship between particle shape and entrainment.”
1254. Minerals Engineering, 83. pp. 211–216.
Lima, N. P., Souza Pinto, T. C., Tavares, A. C., and Sweet, J., 2016, “The Xia, W., 2017, “Role of particle shape in the floatability of mineral
entrainment effect on the performance of iron ore reverse flotation.” particle: an overview of recent advances.” Powder Technology, 317.
Minerals Engineering, 96–97. pp. 53–58. pp. 104–116.