Powder Technology 204 (2010) 1–10

Contents lists available at ScienceDirect

Powder Technology
journal homepage: www.el sevier. com/l ocate/ powtec

Effect of suspension chemistry onto voltage drop: Application to electro-flotation

G. Montes-Atenas a,⁎, F.J. Garcia-Garcia b, R. Mermillod-Blondin c, S. Montes d
a
School of Chemical Engineering and Analytical Sciences, The University of Manchester, P.O. Box 88, Manchester, M60 1QD, United Kingdom
b
Corrosion and Protection Centre, School of Materials, The University of Manchester, P.O. Box 88, Manchester M60 1QD, United Kingdom
c
Université du Québec en Abitibi-Témiscamingue (UQAT), 445, Boulevard de l'Université, Rouyn-Noranda, Québec, Canada J9X 5E4
d
Laboratorio de Fisicoquimica de Superficies, Departamento de Quimica de los Materiales, Facultad de Quimica y Biologia, Universidad de Santiago de Chile,
Av. Libertador Bernardo O'Higgins 3363, Santiago, Chile

a r t i c l e in f o abstract

Article history: Electrochemical processes involving particle suspensions are still ill understood. Electro-flotation is process through which
Received 24 August 2009 particles or ions can be selectively separated. Although the technique has been widely studied, not much is known about the
Received in revised form 4 June 2010 effect of the chemistry of suspended particles on the process performance which constitutes the aim of the present research
Accepted 27 June 2010
work.
Available online 13 July 2010
A column-type cell was designed to carry out electro-flotation experiments and standard air flotation experiments. Two
minerals, Merensky reef and UG2, were tested.
Keywords: Electro- The classic air-flotation operation gave higher recoveries and grade of collected material compared to electro-flotation only
flotation
when using Merensky reef ore. However, in the case of UG2 ore, the electro-flotation reached similar performance to that of
Electrochemistry
the classic operation. It was found that the electrodes readily interact with the mineral pulp reducing the ef ficiency to generate
PGM ore
Suspension bubbles. That efficiency depends significantly on the ore composition. Further electrochemical studies concerning how the
Resistivity voltage is used throughout the process was performed. The conductivity of the pulp formed using UG2 ore increased around 5
Cell design times compared to that observed when using Merensky reef ore. Under identical experimental conditions, the froth formed
with Merensky reef ore was shallow and precarious compared to the thick and stable froth formed in the case of UG2 ore.
© 2010 Elsevier B.V. All rights reserved.

1. When mineral particles are too small, fluid streamlines formed around
Introduction bubbles moving upwards act as an implicit shield for particles travelling
1.1. Historical purpose of electro-flotation process towards the bubble. This “shield” reduces the collision efficiency and it is
considered to be the major reason why the classic froth flotation process is
Originally, the electro-flotation gained some interest since it could solve inefficient at recovering fine particles [10,11].
the problem of recovering fine valuable particles which cannot be separated
using classic flotation operations. Indeed, the froth flotation operation
1.2. Fundamentals of electro-
involving particles 10 μm or less makes the overall selective separation flotation
process completely inefficient [1,2].
The basic principle of the electro-flotation process relies on water
Conventional flotation cells produce large bubbles, ranging in size from
electrolysis process. The reactions taking part can be assumed as follows,
600 to 2000 μm [3]. Some spargers can generate a medium- bubble size (from
100 to 600 μm), but this could be insufficient for ultrafine particle flotation (b
10 μm) [4].
Alternative techniques have been evaluated to accomplish the generation 4OH− → O2 þ 2H2O þ 4e− ð1Þ
of fine bubbles such as dissolved air, electro-flotation, etc. [5–7]. Electro-
flotation has been suggested as the process to obtain the smallest bubble 2H2O− þ 2e− → H2 þ ð2Þ
diameters, ranging in size from 20 to 40 μm [6,7]. Many applications have 2OH−
The half-reaction 1 (oxygen production) is anticipated to occur at the
been developed with this technique such as wastewater treatment to remove
anode while the second (hydrogen production), at the cathode. Both reactions
harmful ions and molecules [8,9].
are described in alkaline conditions normally used in froth flotation operations
linked to sulphide minerals assessed in this work. Venkatachalam (1992) has
performed an exhaustive analysis of the experimental consequences of
◆ Corresponding author. Present address: Julius Kruttschnitt Mineral Research Centre (JKMRC),
producing gas through electrolysis [6].
Australia. Tel.: +61 7 16133655833.
E-mail address: g.montesatenas@uq.edu.au (G. Montes-Atenas).

0032-5910/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2010.06.024
2 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10

During water electrolysis the aqueous speciation or water quality will

change significantly, particularly near the electrodes. Therefore, the first
design decision would be related to how intimately both electrodes and
mineral pulp should be connected. The introduction of membranes or other
partially permeable barriers will act as resistances to mass transfer ultimately
increasing the required applied voltage. If the solution in contact with the
electrodes contains surfactant and/or other reagents commonly used in
flotation opera- tions [12–15], further studies concerning current efficiency
must be performed to evaluate possible electrode passivation or activation.
Concerning the measuring of gas flowrate, it has been reported that the
hydrodynamics plays a significant role [16], therefore a coherent stirring
might be an adequate way to avoid bubbles coalescence. The relationship
between bubble size and current density has not been yet correctly explained.
The bubbles nucleation and subsequent detach- ment from the electrode are
not directly linked to the bubble volume [6]. Yoon and Luttell claimed the
turbulent conditions were unfavourable in conventional mechanical flotation
machines when particles are fine [17]. New machine designs and processes
were proposed such as column flotation, microbubble flotation, agitated
column [18], Jameson cell with external contactor, flotation cyclone, flotation
centrifuge and turbulent microflotation [19].
From a chemical point of view, one of the major advantages of using pure
gases evolving from water electrolysis is their high reactivity. The gas may
activate the mineral surface in order to make improvements in its recovery.
The case of pyrite has been reported to respond well to electro-flotation [20].
However, it is also important to mention that electrolytic gases might also act
against recovery depending on mineral nature [6].
The size of electrolytic bubbles has been found to be pH dependent. In
alkali medium, bubbles of oxygen are bigger than those of hydrogen. Under
acidic conditions, otherwise holds [21].
The main drawback of using electro-flotation is the electric costs. Other
Fig. 1. Column-type cell with two electrodes (SS: stainless steel; Pt: platinum) and an axial stirrer
negative aspects such as the need of expensive cooling system to avoid marine-type mixing impeller. The stirrer speed was set at 800 rpm.
overheating due to charge transfer and the costs linked to the high-quality
electrode material have been also pointed out [22]. Nevertheless, the electrode
material could be chosen in a way to decrease the costs [6]. For example, ores, similarly to Hallimond tube operation. The cell height and diameter are
stainless steel can be used as the cathode. It has also been claimed that the 25 and 5 cm, respectively. The effective volume used of the cell was 400 cm3,
power consumption could be limited by using a current pulsed system that which is around 80% of the total cell volume. The electrode materials were
allows bubble-size control and a reduction of the heat produced [23]. platinum and stainless steel for anode and cathode, respectively. The anode
Additionally, most of the published research on electro-flotation has been and cathode surface areas were 50 and 850 cm2, respectively. The electrolysis
performed using traditional flotation cells [7,22] or even mixing electro- of water was performed by using a power supply capable of delivering
flotation and classic flotation procedures which may mask the real effect of voltages up to 30 V and a maximum current of 1 A.
pure electro-flotation technique. The ores used were Merensky reef and UG2 with grades in sulphur equal
It is important to bear in mind that large scale water electrolysis is not an to 0.24 and 0.014, respectively. The collectors used were SIBX and Senkol 5
untested process. It has been implemented to produce hydrogen; however, the (in mixture), guar gum was used as the depressant and Dowfroth 200 as
industrial conditions used for that purpose are dramat- ically different to those frother. The concentrations used in this work are in agreement with previous
used in minerals recovery by froth flotation [24]. According to our present published literature dealing with the same ores [25–28]. SIBX, Senkol 5 and
knowledge, no formal studies have been reported using electro-flotation for guargum were procured by Senmin (South Africa), while Dowfroth 200 was
platinum group minerals (PGMs). In the present work, the electro-flotation obtained from NASACO International LLC (Switzerland). High purity
process is applied to PGMs, particularly Merensky reef and UG2 ores, sodium hydroxide was used as a pH modifier to ensure enough solution
both belonging to the Bushveld Complex in South Africa. Scanning conductivity.
electron microscopy studies were performed to establish how the different Before assessing the flotation of the ores, the particle size was reduced by
mineral phases included in the ore are associated. Industrial reagents using a rotatory mill achieving the following particle size distributions: 100%
(collectors, depressants and frothers) normally used in plant were used −75 μm (90 min milling) and 75% −75 μm
throughout (70 min milling) for Merensky reef ore; while, 65% − 75 μm
all the experimental work performed at real concentrations. (70 min) for UG2 ore.
The following aspects were considered when designing the cell:

i. The marine-type impeller is used as it produces the desired

2. Experimental
2.1. Materials
A column-type cell (Fig. 1) was built which could work either using air
passing (classic flotation) or electro-flotation. The column- type cell design
was chosen to be able to study the floatability of the
homogeneous solid suspension [29]. The axial movement is dissipated
once the streamlines reach the level of electrodes. Radial movement of
liquid is avoided since that may result in battery behaviour of the
system where the products formed at the anode will be consumed at
the cathode which is undesirable. Classic electrode arrangement used
for example in electrowinning reduces the possibilities to stir the pulp
avoiding particles settling and increasing the interactions between
bubbles and particles. On other hand we allow the electrodes to reduce
the turbulence mid-way up to the lip. In
 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10 3

the upper half of the cell it is desirable to reduce the turbulence and
leave the bubble steadily reach the pulp–froth interface reducing the
chances of particle detachment.
ii. Additionally, it gives the opportunity of using hydrogen and oxygen
simultaneously, avoiding the utilization of further membranes to
separate the gases.
iii. The anode is made of titanium covered by a 1 μm thick layer of
platinum, supplied by ti-shop.com®; whilst the cathode is made from
stainless steel, the same material normally used in industrial water
electrolysis.
2.2. Methods
The reagent concentrations for Merensky reef and UG2 ores used were
different since the grades of these two minerals differ. For Merensky reef and
UG2 ores the concentrations were 40, 18, 12 and 14 g/ton and 50, 50, 100 and
60, for Dowfroth 200, SIBX, Senkol 5 and Guar Gum, respectively.
Two different sets of experiments were performed: (i) conven- tional
flotation and (ii) electro-flotation. The conditions assessed are summarized in
Table 1. The concentration of Dowfroth 200 was increased compared to
previous work, because no froth phase was observed below the value reported
in this work. The times at which the samples were collected were fixed after a
series of preliminary experiments with the column-type cell. All the
experiments were performed with tap water.
It has been reported that the presence of platinum and gold in Merensky
reef is linked to sulphides minerals [30]. Due to low amount of ore used at
laboratory scale, the flotation performance will be reported in terms of the
percentage of sulphur present in the floated fraction.
Therefore, in the case of Merensky reef, the flotation kinetics was
assessed by measuring the sulphur content of the recovery. The
2.2.1. Experiments with air (classic flotation)
In the column-type cell, the air flow rate used varied from around
70 to 180 cm3/min resulting in superficial gas velocities ranging between 0.06
and 0.14 cm/s. The lower air flow rate value was selected because it
approached to the values of gas rate obtained with electro-flotation
experiments. When gas flow rates obtained with air were equal to the gas flow
rate calculated for electro-flotation did not give any recovery.
2.2.2. Electro-flotation experiments
A power supply PSU 130 was used to deliver a voltage between the
electrodes. The voltage was kept constant near to 30 V and the current was
monitored with time. The estimated gas flow rate was obtained from the
current obtained (Table 1).
In order to compare classic and electro-flotation for similar pulp volume
conditions and cell shape a column-type cell was specially designed to work
with both classic air flotation and electro-flotation (Fig. 1). The pulp density
was kept constant at 30%. A mechanical stirrer was introduced from the top
of the cell with the impeller placed close to the cell bottom to provide enough
energy to maintain the ore suspension homogeneous. The stirring speed used
was 800 rpm. There is a strong stirring action near the base of the column-
type cell is reduced up in the column, passing the electrodes, which act as
deflectors of the stream lines formed. Due to the low gas flow rate, it was
difficult to monitor the kinetics of electro-flotation experiments; therefore,
only the final value is reported and compared to the classic flotation
experiments.

2.2.3. Bubble size analysis

The column type cell was connected to a glass cell under an optical
microscope linked to a camera. While performing the electrolysis of a two-
cumulated recoveries and grades were calculated according to phase system at identical reagent concentration considering (30% pulp
Eqs. (3) and (4). density) and stirring conditions used during electro- flotation, the liquid was
pumped from the column cell, into the flat cell and back by a peristaltic pump.
The bubble sizes were statistically studied. The average bubble size, as well
∑i ðmi γiÞt
Rt = ð3Þ as the standard deviation is reported.
MF γF

∑i ðmi γi Þt
Grade t =
∑imit
where the subscript t represents every time interval while mi and γi represent
the mass recovered and its respective percentage of sulphur, respectively. The
subscript F represents the feed. As all experiments were performed in batch,
the mass and grade (MF and γF) are known.
It is important to note that, in the case of UG2 ore, the distribution of
platinum and gold does not obey the latter rule as the valuable metals can be
linked to almost every mineralogical phase in the ore. In this case, the results
reported are only in terms of mass recovery only.
ð4Þ
2.2.4. Electrochemical analysis
The aim of these experiments is to try to get the fundaments of the
electrochemistry connected to the mineral suspension. Mineral electrodes
were built and their electrochemical response was studied. The apparent
surface area of the electrode built to perform the study is
1.1 cm2. Both cathodic and anodic linear voltammograms were obtained using
the potentiostat. The conductivity of tap water and distilled water at pH 11,
using NaOH to raise the pH, was measured and found to be 1397 and 1311
μS/cm, respectively, which represent less than 6% change with respect to tap
water. Therefore, in this part of the study all experiments were performed
with distilled water.

Table 1
Experimental conditions assessed.

Type pH Ores Air flow Superficial Voltage Kinetics Reagent concentration [g/ton]/conditioning time [s]
rate [cc/min] gas velo. [cm/s] applied [V] times [s]
SIBX Senkol Guar Dowfroth

Air-float 11 Merensky 60 0.051 N.A. 180,360,600 18/120 12/120 14/120 40/60

200 0.170 N.A. 120,240,600
UG2 80 0.068 N.A. 180,360,600 50/120 100/120 60/120 50/60
150 0.127 N.A 30,120,600
Electro-flotation Merensky N.A. 0.002a 30 1080 18/120 12/120 14/120 40/60
UG2 N.A. 0.008a 30 1080 50/120 100/120 60/120 50/60
a
Estimation of total gas flow rate from the current obtained.
4 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10
2.3. Analytical equipment
2.3.1. Potentiostat
A Solartron 1280 potentiostat was used to carry out the
voltammetric sweep analyses. A three-electrode cell was used. The
reference electrode was saturated calomel (SCE). Since the two half reaction
occur on different electrode materials, both the cathodic and anodic reactions
where separately assessed using their corresponding working electrode. The
anode, where oxidation reaction occurs, was made of 0.1 μm platinized
titanium mesh electrode supplied by ti- shop® (UK). The cathode was made
of stainless steel, a well proven material for hydrogen production.
2.3.2. Scanning electron microscope
Field Emission Scanning electron microscope ETCbrunel, Zeiss Supra 35
VP was used to analyze the mineral structure of a sample of dried froth
obtained by the electro-flotation method. Both UG2 and Merensky reef ore
were analysed.
2.3.3. Sulphur analysis
The sulphur analysis of the initial ore and the floated fraction were
performed using a Leco SC 32 analyser. The sample is burnt forming SO2
which passes through an infrared detector determining the amount of sulphur.
The latter outcome is later referred to the known initial ore sample obtaining
the percentage of sulphur in the uptake. The measurements were performed in
duplicate and the average is reported. For conditions where the fraction
floated was below 1 g, the experimental procedure was performed several
times to reach a minimum amount of material to perform the assay.
2.4. Prior safety issues concerning the cell design
Probably one of the major concerns about the cell design is the mixing of
oxygen and hydrogen in the same compartment and its safety implications.
Throughout the experiments, the maximum current achieved in the
experiments was around 100 mA. The electro-flotation experi- ments never
went beyond 20 min. The amount of total hydrogen
formed is approximately 10−3 mol (Appendix A). The pressure
exerted by the column of liquid can be neglected and the total
pressure will be close to the atmospheric value. The pressure reached by
bubbles with diameters in the order of 10 μm is 10−3 atm.
In a study by Lewis and Elbe [31] into the explosive limits of gases
found that at atmospheric pressure, the highest possible temperature before the
gas mixture becomes explosive is 560 °C, which was never reached during the
experiments. On the other hand, it may be said that since the bubbles are so
small, the pressure is so high that a critical condition may be accomplished.
The same reference showed that the mixture only becomes explosive at 430
°C when a pressure of 13.2 atm is applied. The pressure values would only be
reached with bubbles of the order of few angstroms which are quite unlikely
to be happening for both gases at simultaneously. All the data refer to a sealed
vessel which is not our case.
It is not the intention of the authors to neglect the potential danger present
in the process once gases leave the solution. Indeed, precautions of keeping an
aerated environment were considered at all time and it will certainly be one
significant subject of research if electro-flotation becomes a reliable alternative
to the current froth flotation processes.
3. Results and
discussion
3.1. Materials characterisation
Fig. 2 shows SEM images of Merensky reef and UG2 ores tested in this work
at particle size distributions of 75% −75 μm and 65% −65 μm, respectively.
EDX analysis was conducted on both ores. Although the low amount in
the sample, sulphides linked to copper, iron and nickel were found which
indicates the presence of pentlandite, pyrrhotite and chalcopyrite which is in
agreement with earlier studies [32]. The oxide phase is rich in silicon and
magnesium, as expected by the presence of pyroxene and feldspar. However,
the sulphur does not appear to be linked to iron which leaves mostly
pyroxenes containing Mg, Ca and Na only. The oxygen distribution appears
to be widespread over the sulphides species. This is in agreement to the
generation of iron oxi-hydroxides appearing at the oxide sulphides surfaces in
all three minerals previously mentioned [33–37].
The presence of these oxidized phases will produce changes in the ore at
both the chemical and physical level which may subsequently involve
changes in its froth flotation behaviour [38]. Merensky reef ore consists
mainly of pyroxene and feldspar (Fig. 2A.1). Sulphide species are mainly
observed on particles below 20 μm wide with well faceted structures (Fig.
2A.2).
Conversely, the EDX image (Fig. 2B.1) indicated that chromite was
present in high amounts within the UG2 ore. However, the chromite structure
was observed to be linked to plagioclase (Fig. 2B.2).
3.2. Flotation experiments
Fig. 3 shows the effect of air flow rate, particle size distribution and type of
gas (air or oxygen and hydrogen mixture) on the recovery and grade of the
Merensky reef mineral case. The recovery increases with both the time and air
flow rate for each particle size distribution assessed.
Furthermore, better results were found when the particle size distribution
contained a higher amount of smaller particles (100% below 75 μm).
In addition, it can be observed that electro-flotation required a
longer time (nearly twice the time measured in air flotation) to reach 2%
recovery, due to the low gas production rate (Fig. 3B); the grade however,
reached a maximum.
Furthermore, the froth phase obtained with electro-flotation is well packed
which promotes entrainment by reducing drainage throughout the froth. In Fig.
4 it can be observed that very few bubbles burst, leaving cavities in the dried
structure which are of the order of 1 to 2 μm diameter (Fig. 4, “a”). This is in
agreement with the order of magnitude reported for bubble diameter
generated by water electrolysis [22]. The latter implies the coalescence does
not play a significant role in these types of froths. Considering the results
presented, classic air flotation is, by far, more effective than electro- flotation,
not only in the rate of separation but also in composition where no significant
effect is obtained using pure oxygen/hydrogen mixture.
The recovery observed was 0.2% and the floated fraction had a
distribution of 78% above 45 μm which confirms the low performance of
electro-flotation in this case.
Fig. 5 shows the results for UG2 ore. The required time for electro-
flotation is again twice that of the classic air flotation. The recoveries
achieved this time, however, are comparable to that of classic air flotation
(Fig. 5A). The froth depth formed was between 2 and 3 time fold the one
observed for Merensky reef ore obtaining better recoveries. This is in
agreement with the results obtained by Feng and Aldrich [43]. Fig. 5A shows
that a recovery of nearly 25% can only be achieved at a fairly high flowrate.
At low flowrate the classic air flotation required nearly four times the volume
of gas to achieve half of the recovery observed for electro-flotation (Fig. 5B).
The recovery observed was around 3% and the floated fraction had a
distribution of 80%w below 45 μm.
All the previous analyses lead to the conclusion that changes in the
ore type lead to differences in the amount of current passing. In the case of
Merensky reef ore, the current was 0.66 A, whilst for the UG2
 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10 5

Fig. 2. Scanning electron microscope (SEM) microphotograph. (A) Merensky reef ore: (A.1) Secondary electrons coupled with image X of sulphur (darker area in the middle top of the figure), oxygen and
iron, 250× magnification, 20 kV. (A.2) Secondary electrons, magnification of (A.1), 1000×. (B) UG2 ore: (B.1) Secondary electrons coupled with image X of chromium, 20 kV, 50× magnification, slightly
darker areas, (B.2) Secondary electrons, magnification of (B.1), 20 kV, 260×.

ore it was 1.23. The current is a function of the voltage applied. The voltage bubbles close to spherical shape were found, there are corrections in the
will be thoroughly analyzed in the electrochemistry section. horizontal axis which were not made due to the distortion generated by the
glass. The latter indicates that only relative changes in bubble size were
3.3. Bubble size analysis considered and not the absolute values.

The bubble size was assessed using an optical microscope connected to
the column-type cell. These experiments were performed in the absence of
mineral but they considered equivalent reagent concentrations.
Fig. 6 shows the pictures of bubbles obtained with tap water in absence of
frother (A) and in presence of 15 mg/L Dowfroth 200 solution (B). A total of
six pictures were taken; the bubbles were counted and measured. The
presence of frother reduced the bubble
size from 99 μm±40 μm to 45 μm±15 μm (Table 2). This means
statistic deviation of around 50% respect to the mean bubble size was found.
Strikingly, the addition of the collectors and depressant slightly increased
the bubble size. The stirring speed did not show a significant effect on the
bubble size measured.
Although the bubble sizes observed are generated at similar conditions
(pH 11, and ambient temperature) during the flotation, the particle size
distributions of the floated fractions are totally different.
3.4. Electrochemistry of the
process
Fig. 7 shows the current density obtained at different separation distances
between the cathode and anode when voltage is adjusted between 0 and 30 V
in a solution of distilled water at pH 11, a 30% weight suspension of
Merensky reef and a 30% suspension of UG2, separately. An almost linear
dependence is observed between the current density and the applied voltage.
As expected, the voltage increased with the distance between electrodes. The
amount of gas evolved becomes significant above a current density of 10
mA/cm2. No significant effect of pulp density was observed in the range of 20
to 30% weight. However, it can be observed that for an equal distance
between electrodes, the current attained with and without the mineral
suspension changed dramatically. Indeed, in general terms, the presence of
UG2 mineral reduces the current density while the Merensky reef increases it.
The cost of the water electrolysis is proportional to voltage applied [24].
The total voltage applied in the cell can be calculated as Eq. (5).

This result implies that the bubble-particle size ratio may not be the only key
factor determining the recovery. Indeed, the chemistry and particularly, the
electrochemistry or rate at which electrodic processes are accomplished is
affecting the recovery. In spite of the fact that ΔEcell = ΔE0 + ηa + jηc j + IRcirc + ð5Þ
IRsol
6 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10
Fig. 3. Cumulative sulphur recovery percent as a function of (A) time and (B) volume of gas
(either air or electrolytic) obtained with modified Hallimond cell. Ore: Merensky reef. Conditions:
two particle size distributions, 70% and 100% below 75 μm and two air flow rates, 60 and 200
cm3/min; electro-flotation, 30 V, 0.65 A, pH= 11. Next to each point the grade of the corresponding
sample is reported.
Where ΔECell is the overall cell voltage, ΔE0 is the standard cell voltage, ηA
is the anodic overvoltage, |ηC| is the cathodic overvoltage, Rcirc represents the
voltage drop due to the installation (electrical
circuitry) and IRso is the potential drop due to the electrolytic solution.
Fig. 4. SEM photograph of the dried froth obtained with Merensky reef ore using electro-flotation, 30
V, 0.65 A, pH 11.
Fig. 5. Cumulative sulphur recovery percentage as a function of (A) time and (B) volume of gas
(either air or electrolytic) obtained with modified Hallimond cell. Ore: UG2. Conditions: particle
size distributions, 60% below 75 μm and two air flow rates, 80 and 150 cm 3/min; electro-flotation,
30 V, 0.9 A, pH= 11. Next to each point the grade of the corresponding sample is reported.
Since the curves obtained with and without the suspension are different,
the only explanation possible is that the mineral particles somehow interact
with the electrodes. The conductivity of the solution after adding the ores was
found to be practically unchanged, 1470 μS/cm. Therefore it can be assumed
that the last term to the right of Eq. (5) will not change significantly.
In the case of many reactions occurring simultaneously, Eq. (5) can be
expressed as Eq. (6).
ΔEcell = Maxi Emix⁎ g + IRcirc + IRsol ð6Þ
f
where Em⁎ix = ΔE0 (a) + ηa = ΔE0(c) + ηc
. The first term on the right hand side takes into account the
maximum current density, resulting from the occurrence of several
simultaneous electrochemical reactions. It will be valuable to get
some idea of the weight of each term on the overall voltage applied. The
strategy will be to calculate the term IRsol for each mineral and estimate the
weight of the solution-electrode interaction term.
Theoretically the solution resistance can be estimated by Eq. (7).
1d
R= ð7Þ
χ
A
Where χ, is the solution conductivity [S/cm], d is the distance
between the electrodes [cm] and A is the surface area [cm2]. The equivalent
conductance of NaOH at the concentration used in the
 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10 7

Fig. 6. Optical microscope pictures of the bubbles generated by using electro-flotation.

Conditions: voltage: 30 V; stirring speed: 800 rpm. (A) Solution at pH 11, without frother. (B)
Solution at pH 11, with Dowfroth 200 at 15 mg/L. The addition of collectors increased slightly the
bubble size observed in (B).

experiment can be calculated by linear interpolation obtaining [38,39] at 18

°C a value of 0.00245 (Ω cm)−1. The conductivity of the alkaline solution was
experimentally determined, giving 0.0015 S/cm, fairly close to the value Fig. 7. Current densities obtained at different lengths of separation between electrodes over a
already reported. However, mineral particles affect voltage ranging from 0 to 30 V. Conditions: pH 11 and pure solution, pH 11 and 30% weight
both, the electrochemical reaction terms and the electrical properties of the of UG2 ore with 60% below 75 μm; and pH 11 and 30% weight of Merensky reef ore with
solution. In order to separate those two effects, a fairly good approximation 70% below 75 μm. (Note: the separation, d, used for electro-flotation in the present work was
4 cm).
would be to realise that the divergence of the curves in Fig. 7, once the
separation of the electrodes is performed, is dependent on the solution
capacity to transport charge. This is particularly true when the applied voltage estimated considering the solution at same pH without the suspend- ing solids
is large. Therefore, considering 2 curves with distances between the as the reference (Eq. (9)).
electrodes equal to d1 and d2, the difference is given in Eq. (8).
/
I d1 −d2
fΔEd1 −ΔEd2 gj
I = ð9Þ
χ
fΔEd1 −ΔEd2 gj I = (IRpulp; −
d1 IRpulp; d2 \ ∀ ΔE = ΔE is large ð8Þ

The only consideration when using Eq. (9) is that the cell geometry should
The term to the left of Eq. (8) can be evaluated using the data plotted in not be changed and in our case it was kept constant for all distances between
Fig. 7. From the term to the right, the conductivity can be the electrodes assessed. The conductivity of double

Table 2
Mean bubble sizes observed during water electrolysis at pH 11.

Summary Solution without Merensky conditions UG2 conditions Merensky conditions UG2 conditions
reagents (low Dowfroth conc.) (high Dowfroth conc.) (all reagents) (all reagents)

Bubble size average [μm] 99.9 45.2 44.3 48.9 56.7

Standard deviation 36.8 15.0 13.0 15.1 26.3
Error % 0.37 0.33 0.29 0.31 0.46
8 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10
Fig. 8. Contributions of solution resistance to the overall voltage applied. Conditions: solution at
pH 11 (NaOH), 30% solids, with same particle size distribution recoveries shown in Figs. 4 and 6
for Merensky reef and UG2 ores, respectively.
distilled water at pH 11 was found to be 0.002 S/cm. The conductivity of the
pulp with Merensky reef and UG2 ores were 0.0136 and 0.0052 S/cm,
respectively.
Fig. 8 shows the percentage corresponding to the solution resistance of the
overall voltage applied. Although most of the values fall in the range of 0 to
60% for the separation of the electrodes assessed in this work, it can be
observed that the shape of percentage vs voltage applied is characteristic for
each mineral. It is interesting to see that in the case of UG2 ore there is one
experimental point falling quite off the trend of the rest of the points. This can
be initially considered as product of the error in the results. The erratic
electrochemical behaviour of UG2 pulps has led us to believe that it might not
only be the error in the measurements. This is addressed in more details
below. However, leaving that point aside the results presented an error below
5%.
3.5. Specific mineral–electrode
interactions
The minerals interact with one or both electrodes, either adsorption or
chemical reaction or both. The latter conclusion can be checked by studying
the electrochemical response of both minerals individually at pH 11. The
results are shown in Fig. 9.
The reactions occurring at the electrodes are very different to those of
only hydrogen and oxygen production. A thermodynamic analysis is
presented in the following paragraphs and subsequently some
hypotheses concerning the kinetics of the whole electro-flotation process will
be proposed.
Considering the Pourbaix diagram [39] the standard potentials associated
to each half-reaction at pH 11 should be calculated from,
(
½OH − 2
1 1
ΔE0 = ΔE0O2−ΔE0H+2 0 0591
=4
: PH=2P
2
O 2
ð10Þ
= 1:229 + 0:1182 Log ½OH−
The values obtained are 0.1 and −0.17 V vs SCE (standard calomel
electrode) for anodic and cathodic, respectively. These values are in
agreement with the results obtained and showed in Fig. 9. In the case
of pure double distilled water, the potentials observed for anodic and cathodic
reactions were 0.366 and −0.430 V vs SCE, respectively. On the contrary, the
values obtained for Merensky reef and UG2 ores were 0 and −0.48 V and;
0.64 and −0.57 V vs SCE, respectively (Fig. 9A).
In the case of Merensky reef ore, reactions arise before hydrogen and
oxygen have even begun to evolve. This is a complicated problem because the
mineral particles approach to the electrode surface and there will be a certain
amount of time where the particles may touch the electrode. That residence
time is in the order of magnitude required to achieve a charge transfer
reaction. The latter will lead to a reduction in the efficiency of the gas
production. Linear voltammo- grams obtained with mineral electrodes reveal
that the open circuit potential (OCP) for the mineral electrodes built are very
far away from the classic condition of zero current for oxygen and hydrogen
evolution (Fig. 9A inset). The y-axis can only be taken as a reference since
the absolute current density will depend on the surface area exposed and the
time where the polarisation is performed. Fig. 9B confirms the fact that both
minerals present different reactivity when linear voltammograms are obtained
with the pulp using the platinum and stainless steel electrodes at higher
applied voltages. The scattered results in the case of UG2 only support the
fact that the current depends on how particles interact with the electrode.
Two main hypotheses may be considered to explain this fact:
i. The mineral particles collide with the electrodes. Since most of the
sulphides have semiconductor properties, their surface attain similar
potential to that exerted by the electrode itself. Conse- quently, the
mineral surface is strongly polarized allowing it to generate the gas. This
path can be considered as a global increase in electrode area.
ii. The mineral particles may be charge carriers. Indeed, once they are
immersed in the aqueous solution, the particles surface is electrically
charged, which is a result of the double layer formation. Those surface
charges can be polarized, spatially deformed and transferred through the
collisions taking place in the suspension. The latter may lead to the
increase of the charge transfer rate through solution. This could be the
case for UG2 where the chromite structure can be stabilized near the
anode which would not be possible for Merensky reef ore due to the
instability observed for purely sulphide species in oxidizing media. The
latter can be reinforced by the fact that the mineralogical composition of
Merensky reef ore consist of feldspar and pyroxene which do not develop
large surface charges [40,41]. In the case of UG2, a significant mineral is
the chromite which does develop significant negative surface charge [42].
These possible theories will be assessed in future work and go beyond the
objective of this paper. However, after determining the values of the
conductivities, the weight of reaction terms can be evaluated and it can be
ascertained whether the mineral–electrode interactions help or hinder the
overall selective separation process.
 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10 9

Fig. 9. (A) Voltammogram obtained using platinum and stainless steel electrodes for anodic and cathodic reactions of (a) double distilled water solution at pH 11 (b) Same as (a) and UG2 ore, 30%; (c)
Same as (a) and Merensky reef ore, 30%; Insert: Voltammograms performed with mineral electrodes made of: (b) UG2 and (c) Merensky ores. (B) Current density vs potential for anodic reaction onto
platinised titanium electrode and cathodic reaction over stainless steel electrode. Conditions: (a) and (a′): pure double distilled water, pH 11;
(b) and (b′) UG2 ore, 30% weight pulp density, pH 11; (c) and (c′) Merensky reef ore, 30% weight pulp density, pH 11. The particle size distribution is the same used to build Figs. 3 and 5. The scan rate
chosen was 10 mV/s.

4. Conclusions polarised electrodes. A stronger interaction was found with Merensky reef ore
in solution which might be connected to the unstable state of pure sulphide
The electro-flotation process proved to be very inefficient when the ore minerals. On the contrary, UG2 ore containing chromite could be much more
grade was relatively high such as the Merensky reef assessed (0.24%); while stable and even help to transfer the current instead of consuming it.
acceptable recoveries and kinetics were achieved when an ore with lower Further studies must be conducted in order to fully understand the
grade, like UG2 (0.014%), was used. The classic froth flotation works much interactions occurring between the mineral and the polarised electrodes, as
better with Merensky reef ore. these may have a positive or negative effect on the outcome.
The voltage required is fairly independent of the pulp density when within
the range of 20 to 30% solid to liquid ratio; however, the distance between the
electrodes is a key issue regarding the low conductivity of a natural media
where froth flotation is usually carried out and constitutes a design variable.
The voltage increases with the current density linearly and it was found that
Acknowledgements
the real potential drop within the solution decreases when the mineral is
added. The main reason for it is that the mineral particles readily interact
with the The authors would like to acknowledge Professor J.J. Cilliers for supplying
the materials to perform this study. Many thanks to Professor
10 G. Montes-Atenas et al. / Powder Technology 204 (2010) 1–10
R.A. Cottis for his invaluable help towards the results and discussion part of the
paper.
Thanks to Mr. John Riley and Mr. Paul Rothwell for all the assistance
provided in performing the experimental work in the best conditions possible.
Appendix
A 66.
According to Faraday Law, the current and the rate at which a reagent is
produced can be computed as,
dM I N.K. Khosla, S. Venkatachalam, P. Somasundaran, Journal of Applied Electrochem- istry 21
= (1991) 986.
dt nF
Where M represents the number of mol (produced or decomposed),
I is the current, n is the stochiometric factor and F is the Faraday constant
(96,500 C/mol). The current is 0.1 A, the time does not go beyond 20 min
(1200 s) generates around 10−3 mol of hydrogen.
References
[1] M. Yekelerand, I. Sonmez, Colloids and Surfaces A: Physicochemical and Engineering
Aspects 121 (1997) 9.
[2] Z. Dai, D. Fornasiero, J. Ralston, Advances in Colloid and Interface Science 85 (2000) 231.
B.A. Wills, Mineral Processing Technology, Butterworth-Heinemann, Oxford, UK, 1997.
[3] R.-H. Yoon, Minerals Engineering 6 (1993) 619.
D.R. Ketkar, R. Mallikarjunan, S. Venkatachalam, International Journal of Mineral Processing 31
[4] (1991) 127.
[5] S. Venkatachalam, Mineral Processing and Extractive Metallurgy Review 8 (1992) 47.
S.E. Burns, S. Yiacoumi, C. Tsouris, Separation and Purification Technology 11 (1997) 221.
A. Khelifa, S. Moulay, A.W. Naceur, Desalination 181 (2005) 27.
[6]
P. Gao, X. Chen, F. Shen, G. Chen, Separation and Purification Technology 43 (2005) 117.
Z.A. Zhou, Z. Xu, J.A. Finch, H. Hu, S.R. Rao, International Journal of Mineral Processing 51
[7]
(1997) 139.
A.V. Nguyen, H.J. Schulze, Colloidal Science of Flotation, 1st ed.Marcel Decker, Switzerland,
[8] NY, 2004.
[9] J. Ralston, D. Fornasiero, N. Mishchuk, Colloids and Surfaces A: Physicochemical and
Engineering Aspects 192 (2001) 39.
[10] S. Montes, G. Montes Atenas, Minerals Engineering 18 (2005) 1032.
S. Ata, N. Ahmed, G.J. Jameson, Minerals Engineering 17 (2004) 897.
[11] W. Yuhua, R. Jianwei, International Journal of Mineral Processing 77 (2005) 116.
[12]
[13]
[14]
[15]
[16] R. Krishna, J.W.A. De Swart, D.E. Hennephof, J. Ellenberger, H.C.J. Hoefsloot, AIChE Journal
40 (1994) 112.
[17] R.-H. Yoon, G.H. Luttrell, Mineral Processing and Extractive Metallurgy Review 5 (1989)
101.
[18] J.A. Finch, G.S. Dobby, Column Flotation, Pergamon Press, Oxford, 1990.
[19] I. Demers, Environmental Synthesis Report, University of Québec in Abitibi-
Témiscamingue, 2005.
[20] K.A. Kydros, G.P. Gallios, K.A. Matis, Journal of Chemical Technology and Biotechnology
59 (1994) 223.
V.A. Glembotsky, A.A. Mamakov, V.N. Sorokina, Electronnaja Obrabotka Materialov 5 (1973)
[21]
H.J. Schulze, Physico-chemical elementary processes in flotation, in: D.W. Fuerstenau, A.J.
[22]
Lynch, J. Laskowski, N.W. Johnson, E.V. Manlaping, C.G. Thorne,
H.J. Schulze (Eds.), An Analysis from the Point of View of Colloidal Science Including
Process Engineering Considerations, 1984, Amsterdam, Oxford, New York, Tokyo.
[23] H. Wuellenweber, J. Mueller, in: S. Srinivasan, F.J. Salzano (Eds.), Proceedings of the
Symposium on Industrial Water Electrolysis, 1978.
[24] Q.I. Wesseldijk, M.A. Reuter, D.J. Bradshaw, P.J. Harris, Minerals Engineering 12 (1999)
1177.
[25] T.D. Mailula, D.J. Bradshaw, P.J. Harris, The Journal of The South African Institute of Mining
and Metallurgy 103 (2003) 143.
[26] J. Wiese, P. Harris, D. Bradshaw, Minerals Engineering 18 (2005) 189.
N. Barbian, Froth stability and flotation performance, PhD Thesis, UMIST — Dept. Chemical
[27] Engineering, Manchester, UK, 2003.
[28] P. Edward, V. Atierno-Obeng, S. Kresta, Handbook of Industrial Mixing — Science and
Practice, John Wiley & Sons, 2004.
I. McDonald, D.A. Holwell, Mineralium Deposita 40 (2005) 526.
[29]
B. Lewis, G. von Elbe, Combustion of Flames and Explosions of Gases, 2nd ed., Academic
Press Inc., NY, London, 1961.
[30]
A.M. Bushwell, D.J. Bradshaw, P.J. Harris, Z. Ekmekci, Minerals Engineering 15 (2002) 395.
[31]
S. Chander, International Journal of Mineral Processing 33 (1991) 121.
P. Velasquez, H. Gomez, J.R. Ramos-Barrado, D. Leinen, Colloids and Surfaces A:
[32] Physicochemical and Engineering Aspects 140 (1998) 369.
Q. Yin, D.J. Vaughan, K.E.R. England, G.H. Kelsall, N.P. Brandon, Journal of The
[33] Electrochemical Society 147 (2000) 2945.
[34] A.M. Buswell, M.J. Nicol, Journal of Applied Electrochemistry 32 (2002) 1321.
D.V. Makarov, W. Forsling, V.N. Makarov, Russian Journal of Electrochemistry 40 (2004) 420.
[35] W. Tolley, D. Kotlyar, R.V. Wagoner, Minerals Engineering 9 (1996) 603.
M. Pourbaix, Atlas d'equilibres electrochimiques, Ed. Gauthier-Villars, France, Paris, 1963.
[36] V. Malysiak, C.T. O'Connor, J. Ralston, A.R. Gerson, L.P. Coetzer, D.J. Bradshaw, International
[37] Journal of Mineral Processing 66 (2002) 89.
N.J. Shackleton, V. Malysiak, C.T. O'Connor, Minerals Engineering 16 (2003) 849.
[38] Q.I. Wesseldijk, M.A. Reuter, D.J. Bradshaw, P.J. Harris, Minerals Engineering 12 (1999)
[39] 1177.
D. Feng, C. Aldrich, International Journal of Mineral Processing 60 (2) (2000) 115.
[40]
[41]
[42]
[43]