Minerals Engineering 77 (2015) 121–130

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

The interaction of clay minerals with gypsum and its effects on

copper–gold flotation
Nestor Cruz a, Yongjun Peng a,b,⇑, Elaine Wightman a, Ning Xu c
a
Julius Kruttschnitt Mineral Research Centre, University of Queensland, Isles Road, Indooroopilly, Brisbane, QLD 4068, Australia
b
School of Chemical Engineering, University of Queensland, St. Lucia, Brisbane, QLD 4072, Australia
c
Minerals and Materials Science & Technology, University of South Australia, Mawson Lakes, SA 5095, Australia

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

Article history: The interaction of two clay minerals, kaolinite and bentonite with gypsum and its effects on the flotation
Received 28 January 2015 of a copper–gold ore was investigated in this study. It was found that bentonite increased the viscosity
Revised 10 March 2015 more than kaolinite when mixed with the copper–gold ore at low shear rates. The detrimental effect
Accepted 11 March 2015
of these clay minerals on flotation was attributed to the entrainment of clay particles when kaolinite
Available online 25 March 2015
was added to the ore and to a decrease in true flotation by bentonite. Bentonite formed a sponge-like
structure with predominant edge–edge (E–E) interactions which might affect hydrodynamics in the flota-
Keywords:
tion cell and have a detrimental effect on flotation recovery. Kaolinite did not form a particular network
Kaolinite
Sodium bentonite
structure and its aggregates mostly consisted of face–face (F–F) type associations which did not affect
Gypsum flotation hydrodynamics. The addition of gypsum to the ore–bentonite mixture inhibited the formation
Rheology of interconnected network structures. This led to lower viscosity values with flotation behaviour similar
Network structures to that of mixtures with kaolinite. In this case, there was an improvement in recovery, but the grade
decreased due to entrainment. The addition of gypsum to the ore–kaolinite mixture created aggregates
with long strings further enhancing particle entrainment with more mass transported to the froth.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The factors influencing hydrodynamics are flotation cell charac-

Clay minerals can be detrimental to mineral flotation and high
viscosity and high particle entrainment promoted by clay minerals
are among the main contributing factors (Forbes et al., 2014;
Jorjani et al., 2011; Wang and Peng, 2014; Wei et al., 2013;
Zhang and Peng, 2015). Incremental reagent consumption has also
been suggested as a factor affecting flotation of ores with a high
clay content (Connelly, 2011). Bentonite has a 2:1 layer structure
and is a swelling clay mineral while kaolinite does not swell and
has a 1:1 structure. Both have a small particle size. Zhang and
Peng (2015) found that bentonite increased the viscosity of a cop-
per–gold ore slurry and decreased copper and gold flotation recov-
ery, while kaolinite affected the viscosity slightly with
simultaneous high gangue entrainment. The decrease in mineral
flotation recovery by bentonite may be via the modification of
hydrodynamic conditions inside flotation cells, and this is related
to fluid flow which is mostly driven by the action of the impeller.
⇑ Corresponding author at: School of Chemical Engineering, University of
Queensland, St. Lucia, Brisbane, QLD 4072, Australia. Tel.: +61 7 3365 7156; fax:
+61 7 3365 3888.
E-mail address: yongjun.peng@uq.edu.au (Y. Peng).
teristics, impeller properties, and slurry properties such as density
and rheology (Shabalala et al., 2011). In this paper the only factor
that is changed is the slurry rheology by adding clay minerals.
This increases pulp viscosity and may cause a high yield stress
and apparent viscosities in the flotation cell, which decreases the
bubble-particle collision efficiency (Schubert, 2008; Xu et al.,
2011) and changes gas dispersion parameters such as bubble size,
gas hold-up, superficial gas velocity and bubble surface area flux
(Shabalala et al., 2011). It is reported that a high yield stress can
form a ‘‘cavern’’ around the impeller (Bakker et al., 2010) and this
decreases bubble size and gas hold-up. When a ‘‘cavern’’ exists
small bubbles are formed in the impeller zone, but the dispersion
of these bubbles is poor in the cell (Shabalala et al., 2011). Patra
et al. (2012a, 2012b) found that network structures formed by
fibrous materials in flotation also interfered with the hydrody-
namic conditions and then decreased mineral flotation recovery.
Their studies suggest that network structures formed by bentonite
may have a similar effect on mineral flotation.
Wang and Peng (2014) studied the negative effect of clay
minerals in flotation through the transport of these particles to
the concentrate. They found that the degree of entrainment of clay
particles was high and also affected by the presence of electrolytes
in the process water. The high particle entrainment in flotation in

http://dx.doi.org/10.1016/j.mineng.2015.03.010
0892-6875/Ó 2015 Elsevier Ltd. All rights reserved.
122 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130
the presence of kaolinite was attributed to the formation of aggre-
gates which entered flotation concentrates and then enhanced
froth stability (Wang and Peng, 2014). Electrolytes in saline water
contributed to the formation of aggregates that were entrained, but
when the clay content was too high, entrainment and true flotation
were affected due to high pulp viscosity which limited bubble and
particle mobility. In that case, edge–edge (E–E) network structures
were enhanced (Wang and Peng, 2014).
Calcium bearing minerals can be present in ores and associate
with clay minerals. In the previous work, it was found that among
the calcium bearing minerals gypsum had the strongest rheological
interaction with bentonite and kaolinite (Cruz et al., 2013).
Gypsum acted as a potential source for Ca2+ and SO24 ions released
into the process water. It is known that multivalent cations are
more strongly attracted to clay particles than monovalent cations
(i.e. Na+ ions) (Lagaly and Dékány, 2013a). Ca2+ cations can increase
the viscosity of kaolinite slurries by attaching to the clay surfaces
modifying the double layers and contributing to the formation of
network structures (Abdi and Wild, 1993; Wild et al., 1993). In
bentonite slurries a high concentration of Ca2+ cations can replace
Na+ cations in the interlayer space deactivating sodium bentonite
such that it is no longer possible for hydration and swelling to
occur (Alther, 1986; Bradshaw et al., 2013; Meer and Benson,
2007). It is also reported that the presence of gypsum and lime
can increase water absorption and swelling pressure of bentonite
(Abdi and Wild, 1993; Wild et al., 1993). The Ca2+ from lime and
gypsum and SO24 anions play a role on this effect.
In this study, the interaction of clay minerals with gypsum in
the formation of network structures and its effects on copper–gold
flotation were investigated.
2. Experimental
2.1. Materials
For the flotation experiments sodium isopropyl xanthate (SIPX),
and the promoter Aero 3894 were used as the collectors. Both
reagents were supplied by Cytec. The frother Polyfroth W22 was
supplied by Huntsman Performance Products, and it is low
molecular weight polyoxyalkylene alkyl ether frother partially
soluble in water. The pH in flotation was adjusted using AR grade
hydrated lime (Ca(OH)2). These reagents are used in the industrial
flotation of the copper–gold ore examined in this study.
The gypsum (calcium sulphate dihydrate) that was added to the
ore–clay mineral mixtures had a high purity (P99%), and was pur-
chased from Sigma–Aldrich. Kaolinite Q38, sodium bentonite, and
quartz were supplied by Sibelco Group, Australia. Q38 is a dry
milled kaolinite with a surface area of 26 m2/g. Quantitative XRD
analysis showed that the composition of the kaolinite sample
was 85 wt.% kaolinite, 4 wt.% quartz and 11 wt.% muscovite, while
the composition of the sodium bentonite sample was 63 wt.%
montmorillonite, 25 wt.% albite and 12 wt.% quartz. The relatively
low-crystallinity of Q38 kaolinite with Hinckley crystallinity index
(IH) of 0.5, derived from the XRD pattern (Plancon et al., 1988), was
chosen in our experiments because it is close to the IH of the
kaolinite in the copper–gold ore (0.6).
The copper–gold ore was obtained from one of the sponsors.
This was considered a low-clay ore. Quantitative XRD analysis indi-
cated that the mineral composition of the ore was 3 wt.% kaolinite,
1 wt.% pyrite, 28 wt.% quartz, 10 wt.% muscovite, 50 wt.% albite
and 8 wt.% clinochlore. The head assays of gold and copper were
0.46 ppm and 0.55 wt.%, respectively. 1 kg crushed ore was ground
to a P80 of 106 lm before preparing the different mixtures
with clay minerals and gypsum. By following this procedure the
valuable mineral liberation was similar for all the experiments.
Particle size measurements were conducted with a Mastersizer
Microplus from Malvern Instruments. This equipment measures
particle size distribution in very dilute solutions by using laser
diffraction with a standard procedure to avoid particle agglomera-
tion. The P80 of both kaolinite and bentonite was about 14 lm, and
the P80 of gypsum was about 77 lm. Further size reduction for
gypsum was difficult due to agglomeration that occurred during
pulverization. Quartz had an original P80 of about 52 lm, and it
was pulverized to achieve a P80 of 14 lm. Its final particle size
distribution was very close to the size distribution of the clay
minerals.
2.2. Preparation of mineral slurries
The low-clay copper–gold ore was used to prepare five different
slurries by mixing it with known amounts of kaolinite (30 wt.%),
sodium bentonite (15 wt.%), kaolinite–gypsum (30–5 wt.%),
sodium bentonite–gypsum (15–5 wt.%), and quartz (30 wt.%). The
low-clay ore with no clay addition was used as a baseline. The
mixture with quartz served two purposes: (1) understanding the
effect of the dilution of copper and gold in the feed on the flotation,
and (2) comparing the effect from quartz and clay minerals. A
30 wt.% quartz content was chosen since it was the highest clay
content used. Although the particle size of the quartz is smaller
than the particle size of the low-clay ore, it can still provide infor-
mation about the effect on recovery when feed grade is changed
during the mixture preparation. The solid concentration for all
the slurries was maintained at 30 wt.% (about 14.6 vol.%) and the
total weight of the mixtures was 1 kg for all cases. The slurries
were prepared using Brisbane tap water.
The low-clay ore was wet ground to a P80 of 106 lm at a
60 wt.% solid concentration. Hydrated lime (800 g/t) and promoter
Aero 3894 (8 g/t) were added to the mill to mimic the plant scale
processing. After grinding, the slurry was transferred to a 3 L
flotation cell where it was well mixed at 1500 RPM and a known
volume of slurry was removed. The same weight of solids that
was removed was replaced with quartz, clay mineral or a mixture
of clay mineral and gypsum. Water was added to make up the
slurry to 30 wt.% for flotation. The removed slurry was filtered,
dried and weighted to confirm that a correct amount of solids
was taken from the original slurry.
After adding the quartz, clay mineral or the mixture of clay min-
eral and gypsum to the flotation cell, stirring continued at 900 RPM
for 30 min when adding quartz or kaolinite and for 45 min when
adding bentonite. When adding bentonite to the ore, agglomerates
were formed and more mixing time was required. It was also
observed that after 30 or 45 min the rheology measurements
became consistent for the mixtures with kaolinite or bentonite.
Once the slurry was mixed, the pH was adjusted to 10 by the
addition of hydrated lime.
2.3. Rheology measurements
A 20 mL syringe was used to draw a sample from the flotation
cell after the addition of flotation reagents. This sample was imme-
diately transferred to the cup of an Anton Paar DSR 301 rheometer.
This cup was part of a Couette geometry (bob and cup) that was
installed for this equipment. A rheogram was produced in 35 s to
reduce the effect of particle settling. The tests were done at ambi-
ent temperature (23 °C), and the shear rate for the rheograms was
between 0.1 and 350 s 1, but the shear rate of the rheograms
shown in this paper was from 0.1 to 222 s 1. For some slurries with
low viscosities, turbulence was created inside the Couette geome-
try at shear rate values more than 222 s 1.

2.4. Flotation experiments
Four concentrates were collected after 1, 3, 7, and 17 min of
flotation. During flotation the impeller speed was set at 800 RPM,
and the air flow rate was 8 L/min. The pH was adjusted to 10 with
hydrated lime before collecting the concentrates and the collector
SIPX and frother Polyfroth W22 were added prior to collecting the
first and third concentrates. The collector dosage was 6 and 4 g/t
before the first and third concentrate respectively. 15 g/t frother
was added before the first and third concentrates. By using this
procedure, a similar flotation performance from the flotation of
the low-clay ore was achieved in the laboratory and plant.
2.5. Cryo-SEM analyses
The Cryo-SEM technique has been used in many applications to
avoid changes in structure induced by vacuum drying or freeze-
drying since the water is vitrified without crystallisation as ice
(Battersby et al., 1994). Two 6 mm long copper tubes with outer-
diameter 4 mm and inner-diameter 3 mm were joined together
by superglue to hold liquid samples. The samples collected from
pulp and the froth of first concentrate were transferred into the
copper tube using a wide-bore pipette. The copper tube with the
sample inside was immediately sealed with dental wax, placed in
a container and plunged into liquid nitrogen. Each sample was
preserved in liquid nitrogen until transferred to the sample
preparation chamber of the PHILIPS XL30 field emission gun
scanning electron microscope (FESEM) equipped with Oxford
Cryo-transfer and fracture stage. The top tube of the two glue-
jointed tubes was then knocked off by a metal knife on the fracture
stage to expose a fresh sample surface inside the tube. The sample
temperature was raised to 175 K to sublime vitrified water at the
rate of 6 nm/s for 2 min with an estimated 720 nm depth of
vitrified water sublimated. The sample was eventually coated by
platinum plasma for 3 min to form a 3 nm thick platinum coating
to avoid charging during the imaging process. FESEM was operated
at 10 kV voltages coupled with Energy-dispersive X-ray spec-
troscopy (EDS) to identify the elemental composition of particles.
A preliminary report on the application of this technique to studies
of kaolinite aggregation in pulp, froth and concentrate phases was
published by Xu et al. (2014).
3. Results
3.1. Flotation
Six flotation tests were conducted using the copper–gold ore
and its mixtures with quartz, clay minerals and gypsum. Fig. 1
100 wt.% ore
30 wt.% kaolinite - 70 wt.% ore
30 wt.% kaolinite - 5 wt.% gypsum - 65 wt.% ore
15 wt.% bentonite - 85 wt.% ore
35 15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
30 wt.% quartz - 70 wt.% ore
30
Mass recovery (%)
25
20
15
10
5 5.00
0 0.00
0 5 10 15 20 25 30
Water recovery (%)
Fig. 1. Mass recovery as a function of water recovery from the flotation of the low-
clay ore and its mixtures with quartz, clay minerals and gypsum.
N. Cruz et al. / Minerals Engineering 77 (2015) 121–130 123
shows the mass recovery as a function of water recovery from
these tests. As can be seen, flotation of the ore or baseline resulted
in a 5% mass recovery and 31% water recovery at the completion of
flotation. The flotation of the ore and its mixture with 30 wt.%
quartz produced very similar mass and water recoveries. The addi-
tion of 15% bentonite reduced the water recovery significantly to
about 9%. This was because a froth phase was almost absent during
the flotation of this mixture. The addition of 30% kaolinite
increased water recovery and mass recovery significantly to 38%
and 24%, respectively. The addition of 5 wt.% gypsum to the
ore–bentonite mixture produced much higher water and mass
recoveries due to the formation of a more stable froth phase. The
addition of 5 wt.% gypsum to the ore–kaolinite mixture also
increased water and mass recoveries, but to a lower degree than
the gypsum addition to the ore–bentonite mixture.
Figs. 2 and 3 show copper grade as a function of copper recov-
ery, and gold grade as a function of gold recovery, respectively,
from the flotation of the ore and its mixtures. The main observation
was the poor flotation outcome that resulted from the mixture
with bentonite and its change after gypsum addition. The baseline
flotation of the ore resulted in a copper recovery of 92% at a grade
of 10% copper (Fig. 2), and 81% gold recovery at a grade of 7 ppm
gold (Fig. 3). These copper and gold recoveries were almost the
same as those obtained from the flotation of the ore–quartz
mixture, but the addition of quartz caused a decrease in grade to
6% for copper and 4 ppm for gold. It seems that in this work the
decrease in feed grade did not affect copper or gold recovery. The
effect on the concentrate grade after the addition of quartz may
be due to the higher entrainment of smaller quartz particle, which
was examined below.
Although copper and gold grades from the flotation of the ore–
bentonite mixture were close to the baseline flotation, copper and
gold recoveries were much lower (Figs. 2 and 3). The addition of
gypsum to the bentonite–ore system changed the flotation
response significantly by improving the copper recovery from
83% to 93% and gold recovery from 64% to 85%. Meanwhile, the
addition of gypsum to the ore–bentonite mixture decreased copper
and gold grades throughout the flotation (Figs. 2 and 3). For exam-
ple, copper grade went down from 8% to 2%, and the gold grade
from about 5 ppm to 1 ppm in the end of flotation. When gypsum
was added to the ore–bentonite mixture, it was observed that the
froth became more abundant and stable as supported by the higher
mass–water recovery curve in Fig. 1.
The addition of kaolinite to the ore did not decrease copper and
gold recoveries from the flotation as the addition of quartz, but did
decrease copper and gold grades to about 2%, and 1 ppm as shown
in Figs. 2 and 3, respectively more than the addition of quartz. The
addition of gypsum to the ore–kaolinite mixture did not further
affect copper and gold flotation significantly.
100 wt.% ore
30 wt.% kaolinite - 70 wt.% ore
30 wt.% kaolinite - 5 wt.% gypsum - 65 wt.% ore
15 wt.% bentonite - 85 wt.% ore
30.0
15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
30 wt.% quartz - 70 wt.% ore
25.0
Copper grade (%)
20.0
15.0
10.0
35 40 45 50 0.00 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Copper recovery (%)
Fig. 2. Copper grade as a function of copper recovery from the flotation of the ore
and its mixtures with quartz, clay minerals and gypsum.
124 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130

100 wt.% ore 100 wt.% ore

30 wt.% kaolinite - 70 wt.% ore 30 wt.% kaolinite - 70 wt.% ore
30 wt.% kaolinite - 5 wt.% gypsum - 65 wt.% ore 30 wt% kaolinite - 5 wt.% gypsum - 65 wt.% ore

Copper recovery by entrainment (%)

20.0 15 wt.% bentonite - 85 wt.% ore 5.0 15 wt.% bentonite - 85 wt.% ore
15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore 15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
30 wt.% quartz - 70 wt.% ore 30 wt.% quartz - 70 wt.% ore
4.0
15.0
Gold grade (ppm)

3.0
10.0
2.0

5.00 1.0

0.0
0.00 0 2 4 6 8 10 12 14 16 18
0.00 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 10.00
Gold recovery (%) Flotaon me (min)

Fig. 3. Gold grade as a function of gold recovery from the flotation of the ore and its Fig. 5. Copper recovery by entrainment from the flotation of the ore and its
mixtures with quartz, clay minerals and gypsum. mixtures with quartz, clay minerals and gypsum.

The increase in recovery can be attributed to recovery by both From these flotation results it can be inferred that the presence
entrainment and true flotation. The Savassi equation (Bıçak et al., of bentonite reduced true flotation, but the presence of gypsum
2012; Savassi et al., 1998) was used to calculate copper recovery caused bentonite to behave more like kaolinite. Kaolinite increased
by true flotation and results are shown in Fig. 4. The overall copper the recovery by entrainment as a result of the creation of a more
recovery by true flotation from the flotation of the ore and its stable froth which is supported by the high mass–water recovery
mixture with quartz was similar (about 92%). The true flotation curve (Fig. 1). This entrainment increased the overall recovery,
recoveries for the mixtures with clay minerals and gypsum were but diluted the concentrate grade (Figs. 2 and 3). The addition of
close to the baseline (about 91%) except for the flotation of gypsum to the kaolinite–ore system promoted additional entrain-
ore–bentonite mixture that produced much lower copper recovery ment, further increasing the recovery but decreasing the grade
by true flotation (82%). However, by examining copper recovery (Figs. 2–5).
from the first concentrate in Fig. 4, all the mixtures apart from
the bentonite mixture, produced higher copper recovery than the
3.2. Rheology measurements
baseline case. It seems that the addition of kaolinite to the ore
increased copper recovery by true flotation marginally, but the
The rheology measurements of the ore and its mixtures with
addition of bentonite decreased copper recovery by true flotation
quartz, clay minerals and gypsum at pH 10 are shown in Fig. 6.
significantly. The addition of gypsum with bentonite altered the
The ore and its mixture with quartz had the lowest shear stress
deleterious effect of bentonite and improved copper flotation by
values and lowest viscosities. The addition of 30 wt.% kaolinite or
true flotation.
15 wt.% sodium bentonite to the ore increased the viscosity as
Fig. 5 shows the copper recovery by entrainment. The flotation
expected (Cruz et al., 2013). The addition of gypsum reduced the
of the ore produced very low copper recovery by entrainment (only
rheological properties of bentonite slightly, but had little effect
0.5%). The addition of quartz to the ore only increased the copper
on the rheological properties of kaolinite as shown in Fig. 6. Due
recovery by entrainment slightly. The flotation of the ore–kaolinite
to the swelling property of bentonite, 15 wt.% bentonite increased
mixture produced a much higher overall copper recovery by
the viscosity to the same degree as 30 wt.% kaolinite when mixed
entrainment (3.2%) than the flotation of the ore–bentonite mixture
with the ore.
(1.2%). The addition of gypsum significantly changed this recovery
Adjusting the pH with hydrated lime added Ca2+ cations to the
especially in the case of the ore–bentonite mixture where the
slurry, and gypsum contributed to additional Ca2+ cations that
recovery by entrainment was increased to 3.0%. In the mixture
should cause kaolinite particles to form agglomerates increasing
with kaolinite, gypsum increased the recovery by entrainment to
the viscosity. However, this increase was not observed when gyp-
3.8%.
sum was added to the ore–kaolinite mixture. The unexpected

100 wt.% ore

100.0 30 wt % kaolinite - 70 wt.% ore
30 wt. % kaolinite - 5 wt.% gypsum - 65 wt.% ore
Copper recovery by true flotaon (%)

2.0 15 wt.% bentonite - 85 wt.% ore

15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
80.0 30 wt.% quartz - 70 wt.% ore

1.5
Shear stress (Pa)

60.0

1.0
100 wt.% ore
40.0 30 wt.% kaolinite - 70 wt.% ore
30 wt% kaolinite - 5 wt.% gypsum - 65 wt.% ore
15 wt.% bentonite - 85 wt.% ore 0.5
20.0 15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
30 wt.% quartz - 70 wt.% ore

0.0
0.00 0 50 100 150 200 250
0 2 4 6 8 10 12 14 16 18 Shear rate (s-1)
Flotaon me (min)
Fig. 6. Rheograms of the ore and its mixtures with quartz, clay minerals and
Fig. 4. Copper recovery by true flotation from the flotation of the ore and its gypsum: total solids concentration 30 wt.% and pH adjusted to 10 with hydrated
mixtures with quartz, clay minerals and gypsum. lime (Ca(OH)2).
 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130
rheology response was further examined by Cryo-SEM analyses in
the next section. In contrast gypsum decreased the viscosity of the
ore–bentonite mixture by either inhibiting the formation of the
bentonite network structures or reducing its swelling properties.
At a first glance the rheograms in Fig. 6 did not show any major
difference between the ore and its mixtures with clay minerals and
gypsum. The rheograms for ore–kaolinite and ore–bentonite
mixtures were similar, but the flotation response for these two
mixtures was very different. One difference between the
ore–bentonite mixture and other mixtures was the higher shear
stress values in the shear rate range between 0.1 and 50 s 1
(Fig. 6). This suggests that for rheology it may be better to compare
the yield stress regions of the rheograms or the shear stress values
at low shear rates. Shear rates in a flotation cell vary from high to
low values depending on the proximity of the slurry to the
flotation cell impeller and the rheology of the slurry (Bakker
et al., 2009). This causes the ore–clay mixtures which are
non-Newtonian, to have different shear stresses and viscosities at
different locations in the flotation cell. It has been demonstrated
that there is a considerable variation of the shear rate distribution
inside the flotation cell with the highest values close to the
impeller (Bakker et al., 2009). The other areas in the flotation cell
may have relatively small shear rate values that can be even close
to zero (Anderson, 2008).
Fig. 7 shows the relationship between apparent viscosity and
shear stress. The ore–bentonite mixture had much higher apparent
viscosity values than the other mixtures in the shear stress range
between 0.5 and 0.9 Pa. In this shear stress range the shear rate
for this mixture is between 0.25 and 6 s 1 (Fig. 6). This means that
for the zones in the flotation cell with shear rates in that range, the
apparent viscosity values are greater when a 15 wt.% bentonite is
present in the mixture with the ore.
Figs. 6 and 7 suggest that the Herschel–Bulkley model may be
more appropriate to calculate the yield stress for the slurries used
in this work since this model fits the yield stress region in the rheo-
grams well. Fig. 8 shows the calculated apparent yield stresses for
the ore and its mixtures with clay minerals and gypsum using the
Herschel–Buckley model. Results show that there is a difference in
yield stress between the ore–bentonite mixture (0.9 Pa) and the
ore–kaolinite mixture (0.5 Pa). The trend in yield stress in Fig. 8
is well correlated to flotation outcomes and the higher the yield
stress, the lower the flotation mass and metal recovery.
3.3. Cryo-SEM analysis
The flotation results confirm that both kaolinite and bentonite
have a detrimental effect on copper and gold flotation, and
2500 100 wt.% ore
30 wt % kaolinite - 70 wt.% ore
30 wt. % kaolinite - 5 wt.% gypsum - 65 wt.% ore
15 wt.% bentonite - 85 wt.% ore
Apparent viscosity (mPa.s)
2000 15 wt.% bentonite - 5 wt.% gypsum - 80 wt.% ore
30 wt.% quartz - 70 wt.% ore
1500
1000
500
0 (Figs. 11b and 12b). These observed changes in network structure
0 0.5 1 1.5
Shear stress (Pa)
Fig. 7. Apparent viscosity as a function of shear stress (Pa) for the ore and its
mixtures with quartz, clay minerals and gypsum: pH adjusted to 10 with hydrated
lime (Ca(OH)2).
125
1.0
0.9
Apparent yield stress (Pa)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Low-clay ore and Kaolinite 30 wt. % - Kaolinite 30 wt % - Bentonite 15 wt.% Bentonite 15 wt.%
Quartz 30 wt.% - Gypsum 5 wt.% - ore 70 wt.% - gypsum 5 wt.% - - ore 85 wt.%
Ore 70 wt.% ore 65 wt.% oren 80 wt.%
mixture
Fig. 8. Calculated apparent yield stresses for the ore and its mixtures with quartz,
clay minerals and gypsum using Herschel–Buckley model.
rheology tests suggest that the high viscosity could be one con-
tributing factor. In this section Cryo-SEM was used to understand
the types of aggregates and network structures formed by clay
minerals and their interactions with gypsum.
Cryo-SEM images in Figs. 9 and 10 compare the type of aggre-
gates and network structures present in the pulp of ore–bentonite
and ore–kaolinite mixtures at different magnifications. It is clear
that there are differences in the kaolinite and bentonite particle
aggregates at the same pH. Figs. 9a and 10a show that the ben-
tonite created a honeycomb-like network structure. This structure
appears to be made of many interconnected cells with pore size at
about 5–10 lm in diameter where E–E and E–F aggregates were
predominant. Figs. 9b and 10b show that the kaolinite platelets
associated mainly in the mode of F–F and E–E as dense elongated
strings and the resulting aggregates were joined to each other
commonly in an E–F manner to form loose clumps. No honey-
comb-like network structure was observed in the pulp of ore–
kaolinite mixture.
Since a greater proportion of kaolinite was added compared to
sodium bentonite, 30 and 15 wt.% respectively, the kaolinite aggre-
gates look more crowded with larger aggregates. Despite this, the
slurry containing sodium bentonite had higher shear stress and
viscosity values at low shear rates as it was explained in the rheol-
ogy section. This suggests that the network structure made by ben-
tonite particles provided much more resistance to flow than the
loose kaolinite aggregates.
Wang and Peng (2014) explained that the entrainment of clay
particles in flotation depended on the type of clay particle associa-
tion, and the recovery of clay minerals increased if they entered the
froth as aggregates increasing froth stability. They stated that
edge–edge (E–E) associations of clay platelets gave higher viscosi-
ties resulting in poor flotation. Face–face (F–F) aggregates are more
compact with denser structures giving lower viscosities. In this
current study, this difference in viscosity caused by the E–E and
F–F aggregates was observed in the mixtures with bentonite (E–E
predominant) and kaolinite (mostly F–F associations).
Figs. 11 and 12 show the Cryo-SEM images of the network
structure formed by ore–bentonite and ore–bentonite–gypsum in
flotation pulp at different magnifications. As can be seen, when
gypsum was added to the ore–bentonite mixture, the extensive
E–E connected long strings were formed without any E–F associa-
tions and the bentonite aggregates were thicker due to more F–F
associations as compared to the ore–bentonite mixture
2 2.5
agree with the findings from Morris and Zbik _ (2009) who added
0.05 M CaCl2 to a 3 wt.% bentonite suspension. Although the
5 wt.% gypsum in the slurries used in this work gave approxi-
mately a 0.016 M concentration of Ca2+ cations, the change in the
network structures were still evident. The presence or absence of
126 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130

Fig. 9. Comparison of the sodium bentonite network structures (a) and kaolinite aggregates (b) in the ore–clay mixtures in flotation pulp: sodium bentonite and kaolinite
concentrations are 15 wt.% and 30 wt.% respectively; 1000 magnification.

Fig. 10. Comparison of the sodium bentonite network structures (a) and kaolinite aggregates (b) in the ore–clay mixtures in flotation pulp: sodium bentonite and kaolinite
concentrations are 15 wt.% and 30 wt.% respectively; 4000 magnification.

Fig. 11. Cryo-SEM images showing the network structure formed by ore–bentonite (a) and ore–bentonite–gypsum (b) mixtures in flotation pulp: 1000 magnification.

network structures was reflected in rheological and flotation per-
formance as shown in the previous sections.
The Ca2+ cations released by gypsum appear to be present in
sufficient quantities to attach to the negative surfaces of the
sodium bentonite particles and change their electrical double
layers. Compression of these layers causes the particles to aggre-
gate mostly in F–F orientations from original three dimensional
honeycomb-like structure in the absence of Ca2+ cations. Ca2+
cations can also interfere with the interlayer swelling of the
sodium bentonite, since the exchange of the Na+ cations in the
interlayer for Ca2+ cations is thermodynamically favourable, and
complete replacement of Na+ by Ca2+ may happen (Bradshaw
et al., 2013; Meer and Benson, 2007). Ca2+ cations in the interlayer
only allow for crystalline swelling with an interlayer separation of
up to 20 Å. On the other hand, hydration of Na+ cations in the inter-
layer can increase that separation to several hundred Angstroms
(Luckham and Rossi, 1999), which is called osmotic swelling. In
addition to this, even small amounts of Ca2+ cations can hold
together the silicate layers, and build up band-type networks.
Large amounts of Ca2+ cations contract these networks forming
compact particle aggregates as illustrated in Fig. 13 (Lagaly and
Dékány, 2013b).
Cryo-SEM images of the flotation froth also show some differ-
ences in bentonite particle aggregation as shown in Figs. 14 and
 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130
Fig. 12. Cryo-SEM images showing the network structure formed by ore–bentonite (a) and ore–bentonite–gypsum (b) mixtures in flotation pulp: 4000 magnification.
15 at two magnifications. In the slurry with sodium bentonite
addition long strings of clay particles were formed from the aggre-
gation at the borders of the bubbles before breakage during sam-
pling, and the addition of gypsum made thinner strings with
more compact aggregates that appeared to be made mostly of E–
E and F–F associations. The concentration or packing of these
aggregates was also reflected in the mass–water recovery and
recovery by entrainment that showed the transport of more solids
into the concentrate (Figs. 1 and 5). The aggregates in ore–ben-
tonite mixture formed less compacted strings in E–E, F–F and E–
F association modes. It is interesting to note that the gel-like
materials were present all over the particles in both the pulp and
froth samples. They contained 6 at.% Na, 18 at.% Al, 41 at.% Si,
19 at.% Ca and 16 at.% S detected by using EDS. Some appeared as
threads between the bentonite strings while some presented like
separate films on the string surfaces or between the strings.
However, no gypsum particle has been found in both pulp and
froth samples. It is assumed that the gypsum reacted with ben-
tonite to form the gel-like material.
Figs. 16 and 17 show Cryo-SEM images for the aggregation of
kaolinite in the pulp of mixtures with and without gypsum at
two magnifications. Fig. 18 shows Cryo-SEM images of froth
agglomerates from the ore–kaolinite mixture and the ore–kaolin-
ite–gypsum mixture. Kaolinite aggregates associated in the E–E,
E–F and F–F modes in both pulp (Fig. 16a) and froth (Fig. 18a) of
the mixture without gypsum with much smaller aggregate parti-
cles as compared to the sample with gypsum addition. The aggre-
gates in the pulp (Fig. 16b) and froth (Fig. 18b) of the mixture with
gypsum were much larger than samples without gypsum. The
thick long strings were associated mostly in F–F and E–E modes
Fig. 13. Schematic showing the aggregation of the clay mineral layers with
increasing attraction: (a) single layers, (b) band-type networks, (c) compact
particles (reprinted from Lagaly and Dékány, 2013b).
127
with many inter-aggregate water sites all over the strings
(Fig. 17b and 18b). Similar to the ore–bentonite–gypsum mixture,
some gel-like strings were observed in the froth sample of ore with
kaolinite and gypsum addition (Fig. 17b) again suggesting that
gypsum took part in the formation of these strings.
The presence of these strings was clearer in the froth images
with a higher magnification (Fig. 18). It seems that in the mixture
containing gypsum, the strings attached to kaolinite agglomerates
increasing the entrainment of clay minerals as observed in the
water-recovery curves (Fig. 1). This also resulted in a decrease in
copper and gold grade (Figs. 2 and 3). EDS was used to determine
the element composition of the strings being 0.3 at.% Na, 0.5 at.%
Mg, 17 at.% Al, 26 at.% Si, 7 at.% S, 2 at.% K and 47 at.% Ca.
However, the result was not accurate because the spot size of beam
was relatively too large (>2 lm) and the gel-like strings were
always associated with the kaolinite aggregates making the accu-
rate measurement of individual gel-like strings difficult in our
experiment set-up. Apparently, the strings were very high in Ca.
It is speculated that the Ca2+ cations from the gypsum interacted
with Si, Al, and Na from the clay minerals forming these structures.
Similar findings were described by Abdi and Wild (1993); Wild
et al. (1993) who explained that lime and gypsum interacted with
kaolinite to produce gel structures.
4. Discussion
The two clay minerals used in this work increased the shear
stress values and viscosities in a similar manner when 15 wt.% ben-
tonite or 30 wt.% kaolinite was mixed with the ore, but there was a
significant difference in the distribution of apparent viscosities at
low shear rates. It seems that this difference in apparent viscosity
was related to the different type of network structures formed by
the clay minerals as shown by the Cryo-SEM images. A lower con-
centration of sodium bentonite particles (15 wt.%) gave higher
shear stresses at low shear rates when compared to a higher con-
centration of kaolinite (30 wt.%) in the slurries. This indicates that
the sodium bentonite aggregates are stronger or more difficult to
break, and create more friction when the slurry starts flowing. At
higher shear rates the mixtures with kaolinite showed slightly lar-
ger shear stress values, and this may be related to the higher
kaolinite concentration in the slurry (30 wt.%). It may be the case
that once the network structures in the mixture with 15 wt.% ben-
tonite were broken, there were less clay mineral particles to create
friction at high shear rates compared with the kaolinite mixture.
The differences between bentonite and kaolinite occur as the result
of differences in the clay mineral properties such as particle struc-
ture, swelling behaviour, surface charge density, and cation
exchange capacity. This study indicates that bentonite aggregates
128 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130

Fig. 14. Cryo-SEM images illustrating the change in froth agglomerates from the ore–bentonite mixture (a) to the ore–bentonite–gypsum mixture (b): 1000 magnification.

Fig. 15. Cryo-SEM images illustrating the change in froth agglomerates from the ore–bentonite mixture (a) to the ore–bentonite–gypsum mixture (b): 4000 magnification.

Fig. 16. Cryo-SEM images showing the kaolinite particle aggregates in the mixtures with ore in the absence (a) and presence of gypsum (b) in flotation pulp: 1000
magnification.

have a greater impact on the hydrodynamics in the flotation cell,
and are more likely to interfere with the transport of minerals to
the froth.
At pH 10 kaolinite particles did not form the sponge-like struc-
ture as seen in the bentonite mixture. Kaolinite aggregates were
more isolated compact agglomerates with predominant face–face
(F–F) interactions. A higher concentration of these aggregates
(30 wt.%) was needed to match the viscosity values of the network
structures formed by bentonite at lower concentration (15 wt.%).
However, similar viscosities did not translate to similar flotation
performance. In this case it is important to look at the distribution
of apparent viscosities especially at the low shear rate values since
these values may be present at various locations in the high vol-
ume of the flotation cell when dealing with non-Newtonian
slurries.
Patra et al. (2010, 2012a and 2012b) investigated the effect of
network structures on flotation by using ores containing fibrous
minerals and systems with nylon fibres. The network structures
formed in their systems are different to the ones containing clay
minerals, but there are some similarities in the way they alter
the hydrodynamics in the flotation cell when sodium bentonite is
present. These researchers found that network structures
 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130
Fig. 17. Cryo-SEM images showing the kaolinite particle aggregates in the mixtures with ore in the absence (a) and presence of gypsum (b) in flotation pulp: 8000
magnification.
Fig. 18. Cryo-SEM images showing the change in froth agglomerates from the ore–kaolinite mixture (a) to the ore–kaolinite–gypsum mixture (b): 8000 magnification.
interfered with bubble dispersion, froth formation, and entrap-
ment of aggregates. The pores in the networks were small
(<20 lm) when compared with the bubble size in the flotation cell
(1–2 mm), meaning that bubbles could not penetrate these pores
in the network structures. This caused poor bubble dispersion,
and hindered bubble percolation affecting froth formation (Patra
et al., 2012a). Similar phenomenon may happen with the sponge-
like structure formed by bentonite in Figs. 10a and 11a. The struc-
tures were not greater than 20 lm, and could affect bubble disper-
sion and movement through the pulp. There was little froth formed
during flotation of the ore containing bentonite, and this could be
attributed to the presence of this ‘‘porous’’ structure. It will be also
reasonable to assume that some valuable minerals could be locked
in the cells of the network structure.
The 30 wt.% kaolinite mixture with the ore did not affect froth
formation. The froth was more stable than in the baseline case
and there was significant entrainment of particles as confirmed
by the mass-recovery data. In this case the overall recovery of gold
and copper was slightly improved (Figs. 3 and 4) due to the
entrainment of particles which at the same time diluted the con-
centrate grade. For this slurry the aggregates did not appear to
be interconnected, and the hydrodynamics in the flotation cell
were not badly affected allowing for entrainment of fine gangue
minerals and the formation of abundant froth.
The addition of gypsum to the bentonite mixture shows that
Ca2+ cations inhibited the formation of the sponge-like structure
resulting in lower viscosities values, and froth formation could
occur. The recovery was improved by both true flotation and
entrainment, but grade decreased. This indicates that the type of
network structure is important in flotation and may be manipu-
lated by addition of electrolytes.
129
The detrimental effect of kaolinite was mostly due to the
entrainment. The highest entrainment occurred when gypsum
was added to the kaolinite mixture with the ore. From the Cryo-
SEM images aggregates were connected by long strings that
appeared to be formed by the presence of gypsum. These aggre-
gates did not increase viscosity (Fig. 6), and their entrainment
resulted in transport of more mass to the froth.
In this study frother and collector additions were maintained
the same with and without the addition of clay minerals and there
appeared that the consumption of reagents by clay minerals did
not have a significant effect on the flotation. This may be associated
with the way clay minerals modified the flotation. Kaolinite and
the mixture of bentonite and gypsum enhanced froth stability
and the consumption of frother by clay minerals may not affect
the frothing in flotation. Bentonite formed a sponge-like structure
with a high pulp viscosity interfering with the hydrodynamics in
flotation, and more or less frother in the pulp may not affect the
flotation performance either. Meanwhile, chalcopyrite, the main
copper sulphide mineral in the ore, displays good collectorless
flotation (Zachwieja et al., 1989). The consumption of collector
by clay minerals in this study may not affect the flotation
significantly.
5. Conclusions
The presence of bentonite and kaolinite resulted in the forma-
tion of different types of aggregates in the mixtures with the ore
at pH 10 adjusted with lime. Bentonite aggregates had a porous
or sponge-like structure with predominant edge–edge (E–E) inter-
actions while kaolinite did not form a particular network structure
130 N. Cruz et al. / Minerals Engineering 77 (2015) 121–130
and its aggregates mostly consisted of face–face (F–F) type associa-
tions. The structures in the mixture with bentonite had higher vis-
cosities across a wider range of low shear stress values which
might affect hydrodynamics in certain regions of the flotation cell
and have a detrimental effect on flotation recovery. Recovery in the
mixture of the ore with kaolinite was not affected, but grade
decreased due to entrainment. The addition of gypsum to the
slurry containing bentonite significantly changed the arrangement
of aggregates by inhibiting the formation of interconnected net-
work structures. This led to lower viscosity values with flotation
behaviour similar to that of mixtures with kaolinite. There was
an improvement in recovery, but grade decreased due to
entrainment.
Acknowledgments
The authors gratefully acknowledge the financial support of this
study from the Australian Research Council, Newmont Mining
Corporation and Newcrest Mining Limited as well as the discussion
and suggestion from Dr. Ronel Kappes at Newmont Mining
Corporation and Dr. David Seaman at Newcrest Mining Limited.
The first author also thanks the scholarship provided by the
University of Queensland.
References
Abdi, M.R., Wild, S., 1993. Sulphate expansion of lime-stabilized kaolinite: I.
Physical characteristics. Clay Miner. 28, 555–567.
Alther, G.R., 1986. The effect of the exchangeable cations on the physico-chemical
properties of Wyoming bentonites. Appl. Clay Sci. 1, 273–284.
Anderson, C.J., 2008. Flotation in a novel oscillatory baffled column. In: Department
of Chemical Engineering. University of Cape Town, Cape Town, p. 249.
Bakker, C.W., Meyer, C.J., Deglon, D.A., 2009. Numerical modelling of non-
Newtonian slurry in a mechanical flotation cell. Miner. Eng. 22 (11), 944–950.
Bakker, C.W., Meyer, C.J., Deglon, D.A., 2010. The development of a cavern model for
mechanical flotation cells. Miner. Eng. 23 (11–13), 968–972.
Battersby, B.J., Sharp, J.C.W., Webb, R.I., Barnes, G.T., 1994. Vitrification of aqueous
suspensions from a controlled environment for electron microscopy: an
improved plunge-cooling device. J. Microsc.-Oxford 176 (2), 110–120.
Bıçak, Ö., Ekmekçi, Z., Can, M., Öztürk, Y., 2012. The effect of water chemistry on
froth stability and surface chemistry of the flotation of a Cu–Zn sulfide ore. Int. J.
Miner. Process. 102–103, 32–37.
Bradshaw, S.L., Benson, C.H., Scalia, J., 2013. Hydration and cation exchange during
subgrade hydration and effect on hydraulic conductivity of geosynthetic clay
liners. J. Geotech. Geoenviron. 139 (4), 526–538.
Connelly, D., 2011. High clay ores: a mineral processing nightmare. Aust. J. Min.,
28–29
Cruz, N., Peng, Y., Farrokhpay, S., Bradshaw, D., 2013. Interactions of clay minerals in
copper–gold flotation: Part 1 – Rheological properties of clay mineral
suspensions in the presence of flotation reagents. Miner. Eng. 50–51, 30–37.
Forbes, E., Davey, K.J., Smith, L., 2014. Decoupling rheology and slime coatings effect
on the natural flotability of chalcopyrite in a clay-rich flotation pulp. Miner. Eng.
56, 136–144.
Jorjani, E., Barkhordari, H.R., Tayebi Khorami, M., Fazeli, A., 2011. Effects of
aluminosilicate minerals on copper–molybdenum flotation from Sarcheshmeh
porphyry ores. Miner. Eng. 24 (8), 754–759.
Lagaly, G., Dékány, I., 2013a. Colloid clay science. In: Bergaya, F., Theng, B.K.G.,
Lagaly, G. (Eds.), Handbook of Clay Science. Elsevier Science, Amsterdam, The
Netherlands, pp. 243–345 (Chapter 8).
Lagaly, G., Dékány, I., 2013b. Colloid Clay. Science 5, 243–345.
Luckham, P.F., Rossi, S., 1999. The colloidal and rheological properties of bentonite
suspensions. Adv. Colloid Interface Sci. 82, 43–92.
Meer, S.R., Benson, C.H., 2007. Hydraulic conductivity of geosynthetic clay liners
exhumed from landfill final covers. J. Geotech. Geoenviron. 133, 550–563.
_
Morris, G.E., Zbik, M.S., 2009. Smectite suspension structural behaviour. Int. J.
Miner. Process. 93 (1), 20–25.
Patra, P., Nagaraj, D.R., Somasundaran, P., 2010. Impact of pulp rheology on selective
recovery of value minerals from ores. In: Singh, R., Das, A., Banerjee, P.K.,
Bhattachalyya, K.K., Goswani, N.G., (Eds.), Mineral Processing Technology,
Jamshedpur, pp. 1223–1231.
Patra, P., Bhambhani, T., Nagaraj, D.R., Somasundaran, P., 2012a. Impact of pulp
rheological behavior on selective separation of Ni minerals from fibrous
serpentine ores. Colloids Surf. A 411, 24–26.
Patra, P., Bhambhani, T., Vasudevan, M., Nagaraj, D.R., Somasundaran, P., 2012b.
Transport of fibrous gangue mineral networks to froth by bubbles in flotation
separation. Int. J. Miner. Process. 104–105, 45–48.
Plancon, A., Giese, R.F., Snyder, R., 1988. The Hinckley index for kaolinites. Clay
Miner. 23, 249–260.
Savassi, O.N., Alexander, D.J., Franzidis, J.P., Manlapig, E.V., 1998. An empirical model
for entrainment in industrial flotation plants. Miner. Eng. 11 (3), 243–256.
Schubert, H., 2008. On the optimization of hydrodynamics in fine particle flotation.
Miner. Eng. 21 (12–14), 930–936.
Shabalala, N.Z.P., Harris, M., Leal Filho, L.S., Deglon, D.A., 2011. Effect of slurry
rheology on gas dispersion in a pilot-scale mechanical flotation cell. Miner. Eng.
24 (13), 1448–1453.
Wang, B., Peng, Y., 2014. The interaction of clay minerals and saline water in coarse
coal flotation. Fuel 134, 326–332.
Wei, R., Peng, Y., Seaman, D., 2013. The interaction of lignosulfonate dispersants and
grinding media in copper–gold flotation from a high clay ore. Miner. Eng. 50–51,
93–98.
Wild, S., Abdi, M.R., Leng-Ward, G., 1993. Sulphate expansion of lime-stabilized
kaolinite- II. Reaction products and expansion. Clay Miner. 28, 569–583.
Xu, D., Ametov, I., Grano, S.R., 2011. Detachment of coarse particles from oscillating
bubbles – The effect of particle contact angle, shape and medium viscosity. Int.
J. Miner. Process. 101 (1–4), 50–57.
Xu, N., Fan, R., Smart, R.S.C., 2014. Cryo-SEM investigation of aggregate structures of
clay minerals in flotation pulp and froth. In: Proc. XXVI International Minerals
Processing Conference, Ch 2 Flotation Fundamentals, 1–11, Santiago, Chile.
Zachwieja, J.B., McCarron, J.J., Walker, G.W., Buckley, A.N., 1989. Correlation
between the surface composition and collectorless flotation of chalcopyrite. J.
Colloid Interface Sci. 132 (2), 462–468.
Zhang, M., Peng, Y., 2015. Effect of clay minerals on pulp rheology and the flotation
of copper and gold minerals. Miner. Eng. 70, 8–13.