71

Table 2.2: The average gas bubble size from various electrodes at different current density
(Ketkar et al., 1991).

Electrodes Current density (A/m2)

125 200 250 300 375
Hydrogen gas bubbles diameter (µm)
Stainless steel plate 34 32 29 26 22
200 mesh 39 35 32 31 28
100 mesh 45 40 38 30 32
60 mesh 49 45 42 40 37
Oxygen gas bubbles diameter (µm)
Platinum Plate 48 - 46 - 42
200 mesh 50 - 45 - 38
 72

Table 2.3 Comparison of different flotation processes with Electro-flotation

Treatment process/ Electro- Dissolved air Impeller Settling/ Foam Reference

Parameter flotation (EF) flotation flotation Sedimentation flotation
(DAF) (IF)

Purification of oily wastewater:

Bubble diameter, 1-30 50-100 0.5-2 - Il’in &

μm Sedashova
, (1999)
Specific power 30-50 50-60 100-150 50-100
consumption,
W/m3

Air consumption - 0.02-0.06 1 -

(m3/m3) water

Chemical IC OC+F OC IC+F

conditioning

Process duration, 10-20 30-40 30-40 100-120

min

Volume of sludge, 0.05-0.1 0.3-0.4 3-5 7-10

%

Oil removal 99-99.5 85-95 60-80 50-70

efficiency, %

SS removal 99-99.5 90-95 85-90 90-95

efficiency, %

Purification of radioactive wastewater:

Process duration, h 0.2-0.25 2-5 0.7-0.8 Il’in &

Kolesniko
Purification 92-95 70-80 85-90 v, (2001)
efficiency, %

Moisture of 92-95 98.5-99.8 92-96.5

sediment, %

Volume of 0.1-0.2 17-18 3.3-3.5

sediment, %

Purification of porcelain and faience industrial sewage:

 73

Process duration, h 0.2-0.4 2.0-7.0 Il’in et al.,

(2002)
Consumption of 0.02-0.04 0.20-0.40
coagulant, g/l

Purification 95-99 70-80

efficiency, %

Moisture of 92.0-95.0 98.5-99.8

sediment, %

Volume of 0.1-0.2 17.0-20.0

sediment, %

Note: SS = Suspended solids; IC = Inorganic coagulants; OC = Organic coagulants; F =

Flocculants
 74

Figure 2.1 The Stern–Grahame model of the electrical double layer (Miettinen et
al, 2010)
 75

Figure 2.2: Geometry for bubble on curved

electrode surface (Sarkar et al , 2010)
 76

Figure 2.3: Schematic diagrams of electrodes arrangement: (a) Monopolar; (b) Bipolar;
(c) Vertical; (d) Horizontal.
 77

Figure 2.4: Conventional electrode arrangement for

electroflotation (Chen, 2004).

.
 78

Figure 2.5: Novel electrode arrangement for electro-flotation (Chen,

2004).
 79

Figure 2.6: Alternative electrode arrangement for

Electro-flotation (Shin, 2003).
 80

Figure 2.7: Combined EC–EF (Chen and Chen, 2010)

 81

CHAPTER 3

ELECTROCHEMICAL MODEL OF ELECTRO-FLOTATION

3 Introduction

Electro-flotation (EF) is an electrochemical technology that is effective in separation of

solids and liquids, such as fine slimes of low grade ores, domestic and industrial waste

water, by means of hydrogen and oxygen gas bubbles generated from water electrolysis.

Bande et al (2008) identified three principal advantages of EF. First, dispersed gas bubbles

formed from electrolysis are finer and more uniform compared to air bubbles in

conventional flotation system. Second, the size and density of electrolytic bubbles can be

controlled by varying current density in the flotation medium, thereby increasing the

probabilities of bubble-particle collision. Third, a specific separation application can be

designed via the selection of appropriate electrode surface and solution conditions to obtain

optimum results. Therefore, its applications to the recovery of fine minerals and oil from

oil/water emulsions have become a topic of recent research (e.g. Sarkar et al, 2010; Xu and

Shang, 2009, Bande et al, 2008). Muller (1992) stated that for the 21st century EF would

be the key electrochemically based technology.

The operating cost is one of the primary concerns for the performance of an EF system.

Operating cost depends on the power consumption that is strongly reliant on the electrolysis

voltage and current. The electric power of an electro-flotation cell can be obtained by

multiplying the cell voltage with current. For a higher current, which is more likely applied
 82

for electro flotation, any small raise of the cell voltage can cause a severe rise in the power

demand. Therefore, the prediction of electrolysis voltage under desired operating

conditions is very important for industrial applications. Although EF has been practiced

since 1900s (Elmore, 1905), a comprehensive electrochemical model to simulate

electrolysis voltage in an electro-flotation cell has not been reported in the literature. A

simple model involving terms of activation overpotential, concentration overpotential and

ohmic overpotential of the solution resistance was proposed by several researchers (e.g.

Scott, 1995). However, the prediction of the unknown terms needs to be addressed more

specifically, such as the activation overpotential and the concentration overpotential.

Moreover the bubble effect on the overall electrical resistance in an electro-flotation cell

needs to be considered. In this study, a simplified analytical model has been derived for

the estimation of electrolysis voltage of an electro-flotation cell based on Nernst equation,

Tafel equation, Fick’s model, and Ohm’s law, considering both the activation overpotential

and concentration overpotential. The effect of bubbles in an electro-flotation cell is also

considered in the theoretical model, which is an enhancement to previous studies (e.g. Chen

et al, 2002a).

The objectives of this study is to derive the theoretical models regarding the electrolysis

voltage required in an electro-flotation process and to verify the simulation results by

comparing with experimental data from both this study as well as data obtained from the

literature.
 83

3.1 Electrochemical model

When an electrolyte solution is brought between two electrodes with opposite electric

charges, a direct current (dc) is passing through from the positive pole (anode) to the

negative pole (cathode), and an electric field is established. As a result of electrolysis

reactions, hydrogen and oxygen gases are formed and liberated at cathode and anode,

respectively, in the form of gas bubbles.

The oxygen evolution reaction at the anode is (Sarkar, 2012)

2H2O – 4e → O2 + 4H+ E°A = +1.23 V (3.1)

The hydrogen evolution reaction at the cathode is (Sarkar, 2012)

4H++ 4e → 2H2 E°C = 0 V (3.2)

The overall electrolysis reaction is expressed as

2H2O → 2H2 + O2 Eeq = +1.23 V (3.3)

The operating current in an electro-flotation cell must exceed the equilibrium potential

difference, anode overpotential, cathode overpotential, and ohmic overpotential of the

solution in order to maintain electrochemical reactions (Scott, 1995), as shown in Eq. (3.4).

Ecell = Eeq + ηa, a + ηa, c + ηa, p + ǀηc, aǀ + ǀηc, cǀ+ ηohm (3.4)
 84

where Ecell is the electrolysis voltage (V); Eeq is the equilibrium potential difference for

water split (V); ηa,a is the anode activation overpotential (V), ηa,c is the anode concentration

overpotential (V), ηa,p is the anode passive overpotential (V), ηc,a is the cathode activation

overpotential (V) and ηc,c is the cathode concentration overpotential (V) and ηohm is the

ohmic overpotential (V). For a new electrode system the passive overpotential is negligible

i.e. ηa,p= 0. The typical voltage components in an electro-flotation cell are shown Figure

3.1.

3.1.1 Equilibrium voltage

The equilibrium voltage of an electro-flotation cell can be expressed by the Nernst equation

(Sarkar, 2012):

At anode:

𝑅𝑇 [𝑃𝑂2 ][𝐻 + ]4
𝐸𝐴 = 𝐸 0𝐴 + 𝑛𝐹 ln { } (3.5)
[𝐻2 𝑂]

At cathode:

𝑅𝑇 [𝐻 + ]2
𝐸𝐶 = E 0 𝐶 + 𝑛𝐹 ln { 𝑃 } (3.6)
𝐻2

where 𝐸 0𝐴 and 𝐸 0 𝐶 are the equilibrium potentials of anode and cathode, respectively,

under standard conditions (T = 298 K, P= 1 atm and [H+] = 1 M (i.e. at pH 0)), R = 8.314

J/K mol, is the ideal gas law constant, T is the absolute temperature, n is the amount of

electrons involved in the reaction (at anode n= 4 and at cathode n= 2) and F is the Faraday

constant (96,485 C/mol).

 85

The equilibrium potential difference between the anode and the cathode is

Eeq = 𝐸𝐴 - 𝐸𝐶 (3.7)

Thus,

𝑅𝑇 (𝑃𝑂2 )[𝐻 + ]4 𝑅𝑇 [𝐻 + ]2
Eeq = [ 𝐸 0𝐴 + 𝑛𝐹 ln { }] - [ 𝐸 0 𝐶 + 𝑛𝐹 ln { 𝑃 }] (3.8)
[𝐻2 𝑂] 𝐻2

3.1.2 Activation overpotential

The activation overpotential is related to the electrode kinetics at the reaction site, which

represents the overpotential incurred due to the activation energy necessary for charge

transfer (Meng et al 2006). The anode activation overpotential (ηa) and cathode activation

overpotential (ηC) can be estimated by applying the Tafel Equation as follows (Chen et al

2002b):

𝜂𝑎,𝑎 = aa +ba ln j (3.9)

𝜂𝑐,𝑎 = ac+bc ln j (3.10)

where, j is the current density (A/m2), aa and ac are the constant of Tafel equation at the

anode and cathode respectively, and ba and bc are Tafel slope of Tafel equation at anode

and cathode, respectively.

 86

3.1.3 Concentration overpotential

The concentration overpotential is caused by the resistance to the transport of reactant

species approaching the reaction site, and the transport of product species leaving the

reaction site (Meng et al 2006). In the present study, the equations of concentration

overpotential in an EF system are derived based on a previous study of Chen et al (2002b).

𝛿𝐶𝑚(𝑥)
In an electrochemical reaction, the mass transport, Jm includes diffusion (-Dm ),
𝛿𝑥

convection (Cm(x) v(x)) and electric migration (tmj), and can be calculated based on

Nernst–Plank equation (Chen et al, 2002b) :

𝛿𝐶𝑚(𝑥)
Jm (x) = -Dm + Cm(x) v(x) + tmj (3.11)
𝛿𝑥

Where Jm (x) is the net flux of a species m, Dm is the diffusion coefficient of species m,

m2/s, Cm(x) is the concentration of species m at distance x, (mol/L), v(x) is the convective

velocity of water flow in the current direction at distance x, m/s, tm is the transport number

of species m and j is the current density (A/m2).

Within the diffusion layer adjacent to the electrode surface, non-reactive ions coming from

electrolyte (e.g. Na+ or SO4-) and other sources (such as water or wastewater) produce a

gradient of concentration and thus cause diffusion current (Chen et al, 2002b). However,

the current from those non-reactive ions are equal but opposite to the migration current at

steady state (Chen et al, 2002b). Therefore the net transport and the net current from those

non-reactive ions are considered to be zero (Chen et al, 2002b). Figure 3.2 illustrates the

concentration variation of reactive ions H+ and OH−near the anode and cathode. As the

concentration of H+ near the anode is relatively high, the total current in the anode
 87

diffusion layer is composed primarily of the migration and diffusion of H+. On the other

hand, as the concentration of OH- near the cathode is much higher than that of H+, the

current comes predominantly from the diffusion and migration of OH−. Near the electrode

surface, the convective flux term (Cm(x) v(x)) can be ignored (Scott, 1995). Moreover, at

a high electrical conductivity, tH+ and tOH− approach zero (Chen et al, 2002b). Therefore

Eq. (3.11) becomes Eq. (3.12), which is widely known as the Fick’s model:

𝛿𝐶𝑚(𝑥)
Jm (x) = -Dm (3.12)
𝛿𝑥

3.1.3.1 Concentration overpotential at anode

Under the assumption that the H+ concentration varies linearly across the whole diffusion

layer at anode,

𝛿𝑐 [𝐶𝐻+ |𝑥=0 ]−[𝐶𝐻+ ]

= (3.13)
𝛿𝑥 𝛿𝑎

Where, CH+|x=0 is the concentration of H+ at the anode surface, CH+ is the bulk concentration

of H+, and δa is the diffusion layer thickness at the anode. At the anode Eq. (3.12) becomes

𝐷𝐻+ [[𝐶𝐻+|𝑥=0 ]−[𝐶𝐻+ ]]

JH+ = (3.14)
𝛿𝑎

Where JH+ is the flux of [H+] (mol/m2 s1), DH+ is the diffusion coefficient of the [H+] (m2/s).

Hence the current density, j, as a function of mass transport flux at the anode can be

expressed as (Chen et al, 2002b):

𝑍.𝐹.𝐷𝐻+ [𝐶𝐻+|𝑥=0 ]−[𝐶𝐻+ ]

j = z. F. JH+ = (3.15)
𝛿𝑎

Where, z is the number of electron in the half reaction.

 88

Usually, at anode, [CH+|x=0]>> [CH+]; hence Eq. (3.15) becomes:

𝑍.𝐹.𝐷𝐻+ [𝐶𝐻+|𝑥=0 ]
j= (3.16)
𝛿𝑎

𝑗𝛿
Thus, [CH+|x=0] = 𝑍.𝐹.𝐷𝑎 (3.17)
𝐻+

The concentration overpotential at anode is:

2.3𝑅𝑇 [𝐶𝐻+ |𝑥=0 ]

𝜂𝑐,𝑎 = ( ) log (3.18)
𝑛𝐹 [𝐶𝐻+ ]

Replacing [CH+|x=0] in Eq. (3.18):

2.3𝑅𝑇 𝑗𝛿𝑎
𝜂𝑐,𝑎 = ( ) log [𝐶 ].𝑍.𝐹.𝐷
(3.19)
𝑛𝐹 𝐻+ 𝐻+

Eq. (3.19) can be used to calculate the concentration overpotential of an electro-flotation

cell at the anode.

3.1.3.2 Concentration Overpotential at cathode

Similarly, assuming the OH- concentration varies linearly across the diffusion layer at the

cathode,

𝛿𝑐 [𝐶𝑂𝐻− |𝑥=0 ]−[𝐶𝑂𝐻− ]

= (3.20)
𝛿𝑥 𝛿𝑐

Where, COH-|x=0 is the concentration of OH- at the cathode surface, COH- is the bulk

concentration of OH-, and δc is the diffusion layer thickness at cathode. At the cathode Eq.

(3.12) becomes
 89

𝐷𝑂𝐻+ [[𝐶𝑂𝐻− |𝑥=0 ]−[𝐶𝑂𝐻− ]

JOH- = (3.21)
𝛿𝑐

Where JOH- is the flux of [OH-] (mol/m2 s1), DOH- is the diffusion coefficient of the [OH-]

(m2/s). Hence the current density, j as a function of mass transport flux at cathode can be

expressed as (Chen et al, 2002b):

𝑍.𝐹.𝐷𝑂𝐻− [[𝐶𝑂𝐻− |𝑥=0 ]−[𝐶𝑂𝐻− ]

j = z. F. JOH- = (3.22)
𝛿𝑐

At the cathode, usually COH-|x=0>> [C OH-] therefore the Eq. (3.22) becomes:

𝑍.𝐹.𝐷𝑂𝐻− [𝐶𝑂𝐻− |𝑥=0 ]

j= (3.23)
𝛿𝑐

𝑗𝛿
Thus [COH-|x=0] = 𝑍.𝐹.𝐷 𝑐 (3.24)
𝑂𝐻−

The concentration overpotential at the cathode (Chen et al, 2002b):

2.3𝑅𝑇 [𝐶𝐻+ ]
𝜂𝑐,𝑐 = ( ) log [𝐶 (3.25)
𝑛𝐹 𝐻1+ |𝑥=0 ]

Where, CH1+|x=0is the concentration of H+ at the cathode surface. [CH1+|x=0] can be

replaced in terms of the ionic product for water, Kw, as follows (Chen et al, 2002b):

𝐾𝑤
[CH1+|x=0] =[𝐶 (3.26)
𝑂𝐻− |𝑥=0 ]

𝐾𝑤 𝑍.𝐹.𝐷𝑂𝐻−
Hence, [CH1+|x=0] = (by replacing [COH-|x=0] in Eq. (3.26) from Eq. (3.24))
𝑗𝛿𝑐

Therefore Eq. (3.25) becomes:

 90

2.3𝑅𝑇 [𝐶𝐻+ ]𝑗𝛿𝑐

𝜂𝑐,𝑐 = ( ) log 𝐾 (3.27)
𝑛𝐹 𝑤 𝑍.𝐹.𝐷𝑂𝐻−

Eq. (3.27) can be used to calculate the concentration overpotential of an electro-flotation

cell at the cathode.

3.1.4 Ohmic overpotential

According to the Ohm’s law, the ohmic overpotential of an electro-flotation cell can be

expressed as:

𝜂𝑜ℎ𝑚 = R. i (3.28)

Where, i and R represent the local current and local electrochemical cell resistance,

respectively. In the absence of bubbles, the cell resistance R can be calculated by Eq. (3.29),

𝑑
R=ρ (3.29)
𝐴

Therefore, Eq. (3.28) becomes

𝑑
𝜂𝑜ℎ𝑚 = ρ i (3.30)
𝐴

Where, ρ is the Electrical Resistivity (Ω∙m), which is a reciprocal to the electrical

conductivity (σ), d is the distance between the anode and cathode (m), A is the surface area

of electrode (m2). In an electro-flotation cell, bubbles are produced, which affect the overall

electrical resistance. Because bubbles have negligible conductivity compared to the

electrolyte, they reduce the conductivity of the system and cause the increase of the ohmic

losses (Cooksey et al., 2008). Several approaches have been suggested to calculate the
 91

modified electrochemical cell resistance Rx due to the effect of bubbles (Cooksey et al,

2008), which can be used in an electro-flotation cell as follows:

1 ℎ
Rx =𝜎𝐴 [1−𝛩 + (𝑑 − ℎ)] (3.31)

1 ℎ (𝑑−ℎ)
Rx =𝐴 [𝜎 + ] (3.32)
𝑥 𝜎

Where, 𝛩 is the area fraction of electrode covered by bubbles (for the simplicity, in the

present model it is assumed that for both anode and cathode, the area fraction of electrode

covered by bubbles are equal to 𝛩, h (= h1+h2) is the thickness of bubble layer as shown in

Figure 3.3, σx is the electrical conductivity of the gas–liquid dispersion and σ is the

electrical conductivity of bubble-free liquid. Several theoretical equations are available to

describe the relationship between σx and σ, of which the most familiar are the Maxwell

equation (Maxwell, 1892):

𝜎𝑥 1−𝜀
= 𝜀 (3.33)
𝜎 1+
2

and the Bruggeman equation (Bruggeman, 1934)

𝜎𝑥
= (1- ε)3/2 (3.34)
𝜎

Where, ε is the volume fraction of bubbles which has a relationship with 𝛩 (Vogt, 2003):

2
ε = 3𝛩 (3.35)

Therefore, by replacing the cell resistance, R by the modified electrochemical cell

resistance Rx; Eq. (3.28) becomes

ηohm = Rxi (3.36)

Now, combining Eqs. (3.8, 3.9, 3.10, 3.19, 3.27 and 3.36); Eq. (3.4) can be rewritten as
 92

𝑅𝑇 (𝑃𝑂2 )[𝐻 + ]4 𝑅𝑇 [𝐻 + ]2
Ecell = 𝐸 0𝐴 + 4𝐹 ln { } - 𝐸0 𝐶 - ln { 𝑃 } + (aa +ba ln j) +
[𝐻2 𝑂] 2𝐹 𝐻2

2.3𝑅𝑇 𝑗𝛿𝑎 2.3𝑅𝑇 [𝐶𝐻+ ]𝑗𝛿𝑐

( ) log [𝐶 ].𝑍.𝐹.𝐷
+ (ac+bc ln j) + ( ) log 𝐾 + (Rx) i (3.37)
𝑛𝐹 𝐻+ 𝐻+ 𝑛𝐹 𝑤 𝑍.𝐹.𝐷𝑂𝐻−

Eq. (3.37) is the proposed model to predict the cell voltage for the electro-flotation Using

j (and i) as the independent variable, the numerical values of corresponding ohmic

overpotential, activation overpotentials, concentration overpotentials, as well as the cell

equilibrium potential can be computed by substituting the measured values into the right-

hand side of the equations. The measured values of Rx can be applied in either Eq. (3.31)

or Eq. (3.32).

3.2 Experimental

3.2.1 Electrode materials

Iridium dioxide coated titanium (Ti) plates (ELTECH System Co), was used as the anode

for the study. Figure 3.4 shows the result of X-ray Photoelectron Spectroscopy (XPS)

analysis of the Ti/IrOx-anode. Tantalum and Iridium were found on surfaces as Ta (V) and

Ir (IV), which reveals that IrO2 was used as conductive precious metal oxides, as electro-

catalysts (i.e. participates in electrochemical reactions) and Ta2O5 was used as a

nonconductive metal oxide as dispersing or stabilizing agent as a coating of Ti. Stainless

steel SS 316 was selected as the cathode for the study. The electrochemical properties of

both electrodes are available in open literature ((Mraz and Krysa, 1994; Kelly et al, 2008).
 93

3.2.2 Solution preparation

In this study, the solution used in the test was prepared by dissolving Na2SO4 in tap water,

which has the conductivity of 210-230 μs/cm. Na2SO4 is added to the solution as a

supporting electrolyte for the sole purpose of adjusting the solution conductivity to 2100

µs/cm. Since the dispersed particles usually do not take part in the electrochemical

reactions (Chen et al, 2002b), the electrochemical behavior of the aqueous solution of the

present study (synthesized with Na2SO4) is expected to be identical to the real

water/wastewater experiment.

3.2.3 Electrodes system

As mentioned earlier, the Ti/IrO2-Ta2O5 anode and SS316-cathode combination were

selected to fabricate electrode modules in the present study. Figure 3.5 shows the test setup

for the overpotential measurement. The potential measurements were made at four

locations as shown in Figure 3.5 (b). The anode-cathode spacing varied from 20 mm to 95

mm. As shown in Fig 3.5, the voltage drop near anode and cathode were measured using

two coated this copper rods installed as close as possible to the electrodes at point 2 and

point 3, respectively. The gap between point 1 and 2 is the same as for point 3 and 4, which

is 0.25mm. Therefore, the anode and cathode overpotentials can be calculated from E12

(voltage between points 1 and 2) and E34 (voltage between points 3 and 4), respectively,

using Eq. (3.38) and Eq. (3.39). Similarly, the experimental ohmic overpotential can be

calculated from E23 (voltage between points 2 and 3) from Eq. (3.40). Therefore, E14

(voltage between points 1 and 4) represents the total experimental cell potential which is

also the summation of E12, E23 and E34. As the copper rod was installed outside of the
 94

diffusion layer (the diffusion layer usually lies between 0.1 mm to 0.001mm, Janssen and

Hoagland, (1970)) there will be a ohmic loss (Eat 0.25 mm) between the electrode and the

copper rod, which is calculated by extrapolating the straight line going through points 2

and 3 as shown in Figure 3.6. The voltage of each cell between the electrodes was

measured by a voltage meter (Fluke 27 Multimeter). The total current for the circuit was

maintained by a DC power supply (HP 6545A).

The experimental anode overpotential of Ti/IrO2-Ta2O5 was measured by the following

relationship:

ηanode = E12 - EA -E (at 0.25 mm) (3.38)

Where, ηanode is the anode overpotential. The Equilibrium potential at anode, EA is

calculated from Eq. (3.5). Similarly, the overpotential of SS 316 cathode was measured by

the following relationship:

ηcathode = E34 - |EC| - E (at 0.25 mm) (3.39)

Where ηcathode is the cathode overpotential, the Equilibrium potential at cathode, Ec is

calculated from Eq. (3.6). The experimental ohmic overpotential was expressed in the

following relationship:

ηohm = E23 + 2 E (at 0.25 mm) (3.40)

 95

3.3 Results and discussions

3.3.1 Effect of bubbles

According to Eq. (3.29), by reducing the distance between electrodes, a lower electrical

resistance can be achieved. However, placing the electrodes too close to each other would

increase the void fraction due to the high density of bubbles and therefore, lead to higher

electrical resistance (Nagai et al., 2003, LeRoy et al., 1979). In the present study, to assess

the effect of inter electrode spacing on the electro-flotation cell voltage, a series of tests

were performed. Figure 3.7 shows the effect of inter-electrode spacing (for 20 mm and 95

mm) on the ohmic overpotentials between experimental data and Eq. (3.30). It is important

to mention that in Eq. (3.30), the ohmic overpotential was calculated without considering

the bubble effect. It was found that for the electrode spacing of 95 mm, the experimental

ohmic overpotentials were within 15% of that predicted by Eq. (3.30). On the other hand,

for the electrode spacing of 20 mm, the experimental ohmic overpotentials were more than

40% of that predicted by Eq. (3.30). This indicates the effect of the void fraction at lower

inter-electrode spacing. Another important observation is that the gap between the

experimental data and results predicted by Eq. (3.30) increases with increasing current

density (A/m2). This is reasonable because at a higher current density, the bubble density

increases, which increases the electrical resistance and hence ohmic overpotential.

The effect of bubbles on the electrical resistance of an electro-flotation unit can be

established by studies of the relationship between resistance and inter-electrode spacing. If

bubbles have no effect, then the resistivity of the electrolyte should be constant and

independent of the inter-electrode spacing (Cooksey et al, 2008). In the present study, it is

found that the resistivity increased with decreasing inter electrode spacing, and became
 96

asymptotic at a spacing approximately 10 to 15 mm (Figure 3.8). Hence it can be assumed

that the bubble layer thickness was approximately between 10 and 15 mm, which are

comparable to the thickness reported in the literature (e.g. 10 mm in Houston et al., 1988;

21 mm in Haupin, 1971). Therefore, electrode covered by bubbles can be estimated by the

Eq. (3.31). The electrode covered by bubbles was calculated to be 42% to 52% with a

standard deviation of 3.6 % for different electrode spacing (20 mm to 95 mm). And the

corresponding bubble fraction between the electrodes was estimated to be 25% to 35% by

Eq. (3.35), which significantly increases the cell resistance and hence overall cell potential.

As mentioned earlier that the bubble coverage and the volume fraction of bubbles change

the conductivity of the electrolyte/air bubble mixture within the bubble layer thickness.

Therefore it is very important to know the reduction of the conductivity, σx due to the gas

component. Most expressions found in the literature related to σx are a function of the

volume fraction of bubbles (ε) as shown in Eqs. (3.33-3.34). Eq. (3.33), known as the

Maxwell equation, is used for equal-sized bubbles, on the other hand Eq. (3.34), known as

the Bruggeman equation, is used for bubbles of unequal size (Cooksey et al, 2008). The

relationship between the resistance (using Maxwell Eq., Bruggeman Eq., Eq. (3.29) and

experiment) and the inter electrode spacing for conductivity of 2100 µs/cm is shown Figure

3.9. It is seen that within the experimental conditions in this study, the resistances are

linearly related to the inter electrode spacing in all cases. It was found that the experimental

resistance was close to the resistance predicted by the Bruggeman equation (within 5%),

which reveals that most of the bubbles are unequal sizes. Moreover the significant variation

between the experimental resistance and the resistance calculated from Eq. (3.29) (without
 97

considering the bubble effect) reveals that in the electro-flotation cell model bubble effect

cannot be ignored.

3.3.2 Simulation

The Tafel slope for Ti/IrO2-Ta2O5 as anode is reported as 76 mV (Mraz and Krysa, 1994).

On the other hand, Tafel slope for SS-316 as cathode is reported as 149 mV (Kelly et al,

2008). The Bubble coverage, 𝛩, was taken as 45 % from experiments as described in the

previous section and the corresponding void fraction (ε) is taken as 30%. The thickness of

bubble layer (h=h1+h2) is taken as 1.2 cm (Figure 3.8). The Diffusivity coefficient for H+

and OH- are available in the literature as 9.3 x 10-5 (cm2/sec) and 5.25 x 10-5 (cm2/sec)

respectively (e.g. Baniasadi et al. 2013). The ion product of water, Kw, is 10-14 (Sawyer et

al, 2003). The Diffusion layer thickness is a function of current density. , which is

approximated as (Janssen and Hoogland, 1970):

δc = -0.012j+11.2 and δa = -0.001j +4.105

The developed model (Eq. 37) for the electro-flotation cell can be validated by comparison

with experimental data. A set of experiments, which were not used to measure the bubble

coverage and bubble fraction, is conducted under various conditions, including different

conductivities (5730 µs/cm to 18,000 µs/cm), inter-electrode spacing (20-300 mm), and

current densities (20-180 A/m2). Figure 3.10 shows a comparison between the predictions

by Eq. (3.37) (the Bruggeman Eq) and experimental results. It is seen that the predictions

are in good agreement with the experimental results.

The Current density and Cell voltage relationship is widely used to validate different types

of electro-chemical cell models, such as electro-coagulation (Chen et al, 2002b) and water
 98

electrolysis for hydrogen production (Choi et al, 2004). Therefore, in the present study, Eq.

(3.37) is used for calculating the relationship between the voltage (V) and current density

(j). The theoretical j-V characteristic of the electro-flotation cell is compared with the

experimental data obtained from Chen et al (2002a). Chen et al (2002a)’s experimental

work on electro-flotation of waste water has been selected because the laboratory setup and

test procedures are clearly reported, which can facilitate the theoretical simulation.

Moreover, the inter electrode gap was reduced to 2 mm, which is quite rare for a

conventional electrode system. In Chen et al (2002a)’s experiments, the electrolysis voltage

dependence on the current density by a Ti based anode (Ti/IrOx-Sb2O5-SnO2) were

measured. As shown in Figure 3.11, the agreement between the simulated electro-flotation

cell potential (using Bruggeman Eq.) and the experimental data is very good. On the other

hand, the gap between the experimental cell potential and the predicted cell potential

without considering the bubble effect (based on Eq. (3.29)) increases with an increase in

the current density. The result reveals that Eq. (3.29) (where no bubble effect is considered)

cannot predict the electro-flotation cell voltage, especially at a low inter-electrode spacing.

3.3.3 Overpotential in electro-flotation cell

The Activation Overpotentials for the anode and cathode were calculated using Eqs. (3.9)

and (3.10), respectively, and shown in Figure 3.12. The Anode activation overpotential is

obviously higher than that of the cathode due to the lower exchange current density of

Ti/IrO2-Ta2O5 anode compared to that of SS 316. It should be noted that the activation loss

in the present study was assumed to occur at the electrode/electrolyte interface. In an actual

electro-flotation cell, the thickness of the electrode does have an effect of the activation

loss (Chan et al, 2001), which is ignored in the present study. Anode concentration
 99

overpotentials and cathode concentration overpotentials are calculated using Eqs. (3.19)

and (3.27) respectively. As seen in Figure (3.13), the cathode concentration overpotential

is higher than that of the anode concentration overpotential, which indicates that the

concentration gradient of OH- in the cathode diffusion layer is higher than that of the H+

gradient in the anode diffusion layer under the pH range (6 to 9) studied in the present

study.

The anode and cathode overpotentials (activation + concentration) were calculated by

using Eqs. (3.38) and (3.39) respectively. As shown in Figures (3.14) and (3.15) the

difference between the experimental overpotential and the theoretical overpotential is up

to 10 % for both anode and cathode. At a low current density (less than 40 A/m2), the

difference between the experimental overpotential and the theoretical overpotential were

almost insignificant which increased with an increase of current density. A similar trend

was observed by Haupin (1971). This may be attributed to the fact that the available

electrode surface area is reduced due to presence of bubbles at higher current density.

Overpotentials at anode and cathode have significant contribution to the total cell voltage.

However the contribution at the anode and cathode, respectively, varies under different

operating conditions. For a current density of 150 A/m2, the anode, cathode and ohmic

overpotentials were calculated by using the model equations for different inter-electrode

spacing (2 to 100 mm) and different electrical conductivities (0.5 ms/cm to 15 ms/cm). It

is found that the anode overpotential contributes in the range of 0.15 % to 20.16 % and the

cathode overpotential contributes in the range of 0.13% to 17.76% of the total electro-

flotation cell voltage. Figure 3.16 shows that the highest contribution of the anode and

cathode overpotentials on the total cell potential observed at low inter-electrode spacing (2
 100

mm) and high conductivity (15 ms/cm). On the contrary, at a high inter electrode spacing

(100 mm) and low conductivity (0.5 ms/cm) the relative contribution of the anode and

cathode overpotential is almost negligible (as low as 0.13%).

3.4 Summary

An electrochemical model is developed to represent the relationship between the total cell

voltage and variables such as the equilibrium potential difference for water split, anode

activation overpotential, anode concentration overpotential, cathode activation

overpotential, cathode concentration overpotential and ohmic overpotential in an electro

flotation process, based on the Nernst equation, Tafel equation, Fick’s model, and Ohm’s

law. The simulation results are compared with experimental data of the present study as

well as data obtained from the literature, and good agreement was found. Therefore, the

model can be used to calculate the total required electrolysis voltage for an electro-flotation

cell under different operating conditions. The effect of bubbles between electrodes on the

electrolysis voltage of electro-flotation is investigated. The Bubble fractions between

anode and cathode were estimated to be 30% to 35%, which significantly increase the cell

resistance and hence the overall cell potential.

The Overpotentials at anode and cathode have significant contribution to the total cell

voltage. However the relative contribution on the anode and cathode varies under different

operating conditions. The anode and cathode overpotentials were measured experimentally

and the results were within 10% of that predicted by the model equations. Among all the

approaches used to measure the cell resistance in the present study, the Bruggeman Eq.

(3.34) is found to be best suited for electro flotation.

 101

3.5 References:

Bande, R.M., Prasad, B., Mishra, I.M., Wasewar, K.L. 2008. Oil field effluent water
treatment for safe disposal by electro flotation. Chem. Eng. J., 137: 503–509.

Baniasadi, E. I. Dincer, G.F. Naterer. 2013 Oxygen evolving reactor overpotentials and
ion diffusion in photo-catalytic and electro-catalytic hydrogen production
International Journal of Hydrogen Energy, Volume 38, Issue 14, 10 May 2013, Pages
6112–6119

Bruggeman, D.A.G. 1984. Ann. Physik (Leipzig) 24 1934. 626.

Chan S.H. et al. 2001, Journal of Power Sources 93 .130-140

Chen, X.M., Chen, G., Yue, P.L. 2002a. Novel electrode system for electro-flotation of
wastewaters, Environ. Sci. Technol. 36(4), 778–783.

Chen, X.M.; Chen, G.H; Yue, P.L. 2002b. Investigation on the electrolysis voltage of
electrocoagulation. Chemical engineering science, 57 (2002) pp2449-2455.

Choi, P., Dmitri G. Bessarabov, D.G., Datta, R. 2004. A simple model for solid polymer
electrolyte (SPE) water electrolysis. Solid State Ionics 175 .535–539

Cooksey, Mark A., Mark P. Taylor, and John J.J. Chen, 2008. Resistance Due to Gas
Bubbles in Aluminum Reduction Cells, www.tms.org/jom.html

Elmore, F.E., 1905. A process for separating certain constituents of subdivided ores and
like substances, and apparatus therefore, British patent 13,578.

Haupin, W. 1971. JOM, 23 (10) pp. 46–49.

Houston G.J. et al. 1988. Light Metals, ed. L.G. Boxall (Warrendale, PA: TMS, 1988), pp.
641–645.

Janssen, L. J. J. and Hoogland, J. G. 1970. The effect of electrolytically evolved gas

bubbles on the thickness of the diffusion layer, Electrochemica Acta, Vol. 15. pp.
1013 to 1023. Pereamon Press.

Kelly, R. G. B. Tribollet, F. J. Presuel-Moreno 2008. Modeling and Simulation of

Dissolution and Corrosion Processes, Issue 12, ECS transaction

LeRoy, R. L., M. B. I. Janjua, R. Renaud and U. Leuenberger. 1979. J. Electrochem. Soc.,

126. 1674.

Maxwell J.C, 1892..A Treatise on Electricity and Magnetism’, 3rd edn, Vol. 1 (Clarendon
Press, Oxford 1892), p. 440; 2nd edn, Vol. 1 (Clarendon Press, Oxford, 1881), p. 435.
 102

Meng Ni, Michael K. H. Leung, and Dennis Y. C. Leung. 2006. An Electrochemical Model
of a SolidOxide Steam Electrolyzer for Hydrogen Production, Chem. Eng. Technol.
29, No. 5

Mraz, R., Krysa, J.1994. Long service life IrO2/Ta2O5 electrodesfor electro flotation, J.
Appl. Electrochem. 24, 1262– 1266

Muller, K. 1992. Electro-flotation from the Double Layer to Trouble Waters. In:
Electrochemistry in Transition, Conway, B.E.; Murphy, O.J.; Srinivasan, S., eds.;
Plenum: New York, 21–37.

Nagai, M. Takeuchi, T. Kimura and T. Oka. 2003. Int J Hydrogen Energy, 28, 35.

Sarkar, M.S.K.A., .Evans, G. M., Donne, S. W. 2010. Bubble size measurement in electro
flotation. Minerals Engineering 23, 1058–1065

Sarkar, S.K.A .2012. Electroflotation: its application to water treatment and mineral
processing. PhD thesis, chemical engineering dept. The University of Newcastle.

Sawyer, Clair Perry McCarty, Gene Parkin, 2003. Chemistry for Environmental
Engineering and Science, McGraw-Hill Education, - Science - 752 pages

Scott, K. 1995. Electrochemical Processes for Clean Technology. Royal Society of

Chemistry Cambridge.

Vogt, H. and H.-D. Kleinschrodt. 2003. Ohmic interelectrode voltage drop in alumina
reduction cells, Journal of Applied Electrochemistry 33: 563–569.

Xu, Y, and Shang, J. Q. 2009. Electrokinetic Flotation of Process Water from Paint Booths.
Water Quality Research Journal of Canada, 44 (2), 183-188.
 103

FIGURES

Figure 3.1: A typical potential distribution between electrodes

 104

Figure 3.2 Concentration variation of H+ and OH- near Anode (left) and Cathode
(right) (based on (Chen et al, 2002b))
 105

Figure 3.3: A simplistic model of bubble layer in the electro-flotation cell

between anode and cathode and anode.
 106

Plate Ir 4f/32
2
x 10

Name Pos. FWHM L.Sh. Area %Area

Ta 5s 64.68 4.00 GL(30) 2497.1 0.0
Ir 4f5/2 Ir(IV) Oxide 64.57 1.91 GL(30) 3252.9 0.0
40 Ir 4f7/2 Ir(IV) Oxide 61.64 1.60 GL(30) 4337.3 100.0

35

30
CPS

25

20

15

70 68 66 64 62 60 58 56 54
Binding Energy (eV)
Surface Science W estern

Figure 3.4 XPS analysis Ti plate

 107

Figure 3.5: Voltage measurement points (a) photo (b) schematic diagram
 108

Figure 3.6: Calculation of experimental anode, cathode and ohmic over-potential

 109

Experimental data (20 mm) Experimental data (95 mm) Eq. (30) (20 mm) Eq. (30) (95 mm)

120

100

80
Ohmic overpotential, V

60

40

20

0
0 50 100 150 200 250
Current Density, A/m2

Figure 3.7 Effect of bubbles on ohmic overpotential as a function of current density

 110

16

Approximate bubble
14 layer thickness

12
Resistivity (ohm-m)

10

4
0 10 20 30 40 50 60 70 80 90 100
Spacing (mm)

Figure 3.8 Effect on bubbles on resistivity as a function of inter-electrode spacing

 111

160

140

120
Resistance, R (ohm)

100

Experiment
80
Eq. 29

60 Maxwell
Burgman
40

20

0
0 20 40 60 80 100 120
Inter-electrode spacing, mm

Figure 3.9 A comparison of equations to predict resistance of electro-flotation cell as a

function of inter-electrode spacing
 112

35

30

25
Predicted Cell Voltage, V

20

15

10

0
0 5 10 15 20 25 30 35
Experimental Cell voltage, V

Figure 3.10 Comparison between predictions by Equation 3 and the experimental values
 113

Experimental data (Chen, 2002a)

Model Eq (without considering bubble effect)
Model Eq (considering bubble effect)
25
Electro-flotation cell voltage, V

20

15

10

0
0 50 100 150 200 250 300 350 400 450
Current density, A/m2

Figure 3.11 Comparison between theoretical and experimental results

 114

Anode Cathode

0.4
0.38
0.36
Activation Overpotential, V

0.34
0.32
0.3
0.28
0.26
0.24
0.22
0.2
0 50 100 150 200
Current Density, A/m2

Figure 3.12 Activation over-potential versus current density

 115

Anode Cathode

0.12

Concentration Overpotential, V 0.11

0.1

0.09

0.08

0.07

0.06
0 50 100 150 200
Current Density, A/m2

Figure 3.13 Concentration over-potential versus current density

 116

Experimental Theoretical

0.65

0.6

0.55
Anode Overpotential, V

0.5

0.45

0.4

0.35

0.3

0.25

0.2
0 20 40 60 80 100 120 140 160 180 200
Current Density, A/m2

Figure 3.14 Comparison between theoretical and experimental anode overpotential

 117

Experimental Theoretical

0.55

0.5

0.45
Cathode Overpotential, V

0.4

0.35

0.3

0.25

0.2
0 20 40 60 80 100 120 140 160 180 200
Current Density, A/m2

Figure 3.15 Comparison between theoretical and experimental cathode overpotential

 118

2 mm spacing and 15 ms/cm coductivity 100 mm spacing and 0.5 ms/cm

Relative Contribution in total Cell voltage (%) 120

99.30%
100

80

60 53.39%

40

20.16% 17.77%
20
8.68%

0.15% 0.13% 0.42%

0
Anode Overpotential Cathode Ohmic potential drop Eeq
Overpotential

Figure 3.16 Relative contributions of different components on total cell voltage

 119

CHAPTER 4

BUBBLE SIZE DISTRIBUTION IN LABORATORY SCALE ELECTRO-

FLOTATION STUDY

4 Introduction

Flotation is a unit operation for solid-liquid separation by introducing fine gas bubbles into

the liquid phase. Its separation efficiency is largely dependent on the size of the bubbles

formed. Bennet et al. (1958) first reported the effect of bubble size and found that smaller

bubbles were more effective in the flotation of coal particles. Other studies also confirmed

the role of bubble size in fine particle flotation (e.g. Ahmed and Jameson, 1985; Lee, 1969).

Sarker et al (2010) stated that the conventional mechanically agitated and sparged column

flotation cells produce air bubbles that are too large for the flotation of fine particles.

Ahmed and Jameson (1985) found that bubble size has a strong influence on the flotation

rate constant (K). The first order kinetic constant varies as a function of reciprocal of bubble

diameters (db): db-3 at quiescent conditions (Yoon, 2000) and approximately vary as db-1.5

under turbulent conditions (Heiskanen, 2000). Electro-flotation (EF) has the advantage of

generating finely dispersed hydrogen and oxygen bubbles (Bande et al, 2008). Moreover,

the bubble sizes and concentration can be controlled by the current density, hence enhance

the probabilities of bubble-particle collisions. Electrochemical reactions due to electrolysis

are well understood, i.e. (Sarkar, 2012):

2H2O → O2 + 4H++ 4e (4.1)

 120

4H++ 4e → 2H2 (4.2)

The physical procedure of bubbles formation can be divided into three phases: nucleation,

growth and detachment (Vogt, 1983). When the molecular oxygen and hydrogen form (as

shown in Eqs. 4.1 and 4.2), they moved away from the anode and cathode respectively.

Consequently, an excessive concentration gradient of dissolved oxygen and hydrogen

builds up near anode and cathode, respectively. If the concentration gradient exceeds a

threshold, nuclei at the electrode surface become active, and oxygen and hydrogen are

transformed into the gaseous phase near the anode and cathode, respectively (Vogt, 1983).

Gas bubbles grow by the internal pressure of the bubble (Lumanauw, 2000) as well as mass

transfer of dissolved oxygen or hydrogen from the electrodes to the nuclei of the

corresponding gas. Five different forces act on a growing bubble. Among them, drag force

(Fd) and surface force (Fs) hold the bubble on the electrode surface, whereas liquid inertia

force (Fi), pressure force (Fp), and buoyancy force (Fb) pull the bubble away the surface

(Lumanauw, 2000). When the sum of Fi, Fp and Fb exceeds Fd and Fs, bubbles depart from

the electrode. The bubbles continue to absorb gas from the supersaturated water when

rising from the electrode after being detached from the electrode (Vogt, 1983) and reaches

its maximum size at around 1 mm from the electrode surface (Sarkar et al, 2010). Many

researchers worked on determination of bubble size at the time of departure (e.g. AI-Hayes

and Winterton, 1981), however no satisfactory theory has yet been developed (Lumanauw,

2000).
 121

Bennett et al. (1958) showed that the flotation rate can be increased not only by reducing

the bubble size but also by generating more bubbles. The higher bubble flux provides more

opportunity for collisions. In an EF process, the bubble fluxes, i.e. the oxygen and hydrogen

can be controlled by the current according to the Faraday’s law (Chen and Chen, 2010):

𝐼𝑉𝑜
QO = (4.3)
𝐹𝑛𝑜

𝐼𝑉
QH =𝐹𝑛𝑜 (4.4)
𝐻

Where, QO (L/sec) and QH (L/sec) are the generating rates of O2 and H2 respectively, at the

normal state,V0 is the molar volume of gases at the normal state (22.4L/mol); F is the

Faraday’s constant (96,500 C/mol of electrons); I is the current (Amp), nO is the electrons

transfer number of O2 (4 mol electrons per mole of O2) and nH is the electrons transfer

number of H2 (2 mol electrons per mole of H2). Substituting the values of V0, F, nO and nH;

Eq. (4.3) and Eq (4.4) can be re-written as (Chen and Chen, 2010):

QO = 0.58* 10-4 I (4.5)

QH = 1.16 * 10-4 I (4.6)

Total number of oxygen and hydrogen bubbles generate in an EF cell per unit time can be

approximated by Eq. (4.7) and Eq. (4.8) respectively:

𝑄
NO = fO𝜋𝑑3 𝑂 /6 (4.7)
𝑏/𝑂

𝑄
NH= fH𝜋𝑑3 𝐻 /6 (4.8)
𝑏/𝐻
 122

Where, db/O and db/H are the mean bubble dia of O2 and H2 respectively, fO and fH are the

fraction of total O2 and H2 transformed into bubbles. The procedure for determination of

fO and fH has been described in detail elsewhere (e.g. Vogt, 1984).

The size of the desired bubble generation can be controlled by the current density, pH and

by choosing various metal electrodes with various surface geometries (Raju and

Khangaonkar, 1984). Although bubble size is a key parameter in EF, bubble size has rarely

been studied in lab scale flotation cells. Among the limited number of studies, there is still

contradiction among the researchers regarding the effect of current density on bubble size.

For example, Ketar et al (1991) and Khosla et al. (1991) showed that with an increase in

the current density, the bubble diameter decreases. However, Sides (1986) and Landolt et

al. (1970) reported an opposite effect due to the bubble coalescence at higher current

densities. The study of Burns et al. (1997) suggested that there is no clear trend of bubble

diameters as a function of current density at low values (40- 210 A/m2). Lumanauw (2000)

concluded that the mean bubble size increases with current density for smooth surface

electrodes, whereas the opposite trend is observed for rough-surface electrodes (Liuyi et

al., 2014). It is understandable that the inconsistency among the researches on measuring

the bubble size in open literature is a result of different testing methods, electrode materials,

surface areas and smoothness, as well as the medium conditions such as electrolyte and

pH.

The dimensionally stable anodes (DSA) invented by Beer (1972) in the late 1960s are the

most important anodes in EF due to their corrosion resistance properties and long service

life. However, the effect of DSA anode on size of oxygen bubble has not yet been

investigated. Furthermore, there is still contradiction among the researchers regarding the
 123

bubble size distribution in EF. Burn et al (1997) assumed the normal distribution to

calculate the bubble sample size. On the other hand, Fukui and Yuu (1985) expressed the

bubble size distribution as the log normal distribution.

The uncertainty in influencing factors mentioned above has made it difficult to effectively

design efficient EF systems for fine particle recovery (Sarkar, 2010). This study was aimed

at reducing the uncertainty by investigating the size of hydrogen and oxygen bubbles

produced from Stainless steel (SS 316) and Ti-IrO2 (DSA) electrodes as a function of

current density, pH, electrode geometries, KNO3 concentration and frother concentration.

Furthermore, four mathematical distributions (normal, log-normal, Weibull (Rosin–

Rammler) and gamma distributions) are fitted to the experimental data to assess the

mathematic distributions of bubble sizes in an EF cell.

4.1 Experimental setup

In this study, the qualitative and quantitative observations of oxygen and hydrogen bubble

evolution in EF have are made using image analysis. The experimental setup was consisted

of an electro flotation cell, made of Plexi glass, a DC power supply, a light source, and a

high-speed camera, as shown in Figure 4.1. Iridium dioxide coated titanium plate (TP),

known as DSA, was used as the anode and stainless steel SS 316 plate (SP) was selected

as the cathode. In a few experiments, stainless steel mesh (SM) with wire dia of 1.6 mm

and aperture of 6.87 mm was also used to investigate the effect of the geometry on bubble

size. Details of the electrode configuration as well as experimental conditions are listed in

Table 4.1. The electrodes were attached near the bottom of the cell, with a separation of 15

cm between the anode and cathode, and connected to a DC power supply. KNO3 was used
 124

as the supporting electrolyte to provide ions for conduction. Once the power was turned

on, hydrogen and oxygen bubbles were produced from the cathode and anode, respectively.

The hydrogen and oxygen bubbles were observed through the camera lens, which was

placed (see Figure 4.1) in such a way that it is focused on the bottom electrode. Therefore,

there was no mixing of H2 and O2 bubbles during capturing photos. Images of the bubbles

were taken using an EOS Canon 60D with high-resolution (5184 pixels × 3456 pixels).

Magnifications were made using a Raynox DCR-250 macro lens in conjunction with 18 to

135 mm of Canon Zoom lens. The bubbles were illuminated from behind using a LED light

source. The shutter speed of the camera was set to 1600 per second, with f-number 5.6, to

provide a clear image of fast moving bubbles. The bubble evolution images was taken at

various conditions including different current densities (10 to 150 A/m2), water pH (2 to

12), electrode materials (either steel or Ti/IrO2 as anodes), KNO3 concentration (0.1 M to

0.5 M) and frother (tennafroth 250) concentration (10 to 30 mg/L). The resolution of each

image was calibrated by taking photo of a scale located in the plane of the rising bubbles.

The images were examined using Imagej software to obtain the bubble diameter. All

experiments were performed using de-ionized water. A typical image of the bubbles is

shown in Figure 4.2.

4.2 Bubble size distribution

The size distribution of bubbles is the key in flotation as only bubbles of certain size serve

a useful purpose (Ben-Yosef et al, 1974). Therefore, it is the key to know the exact bubble

size distribution in designing a flotation cell. Tavlarides and Stamatoudis (1981) and Pacek

et al. (1998) reported that the log-normal distribution describes droplet size distributions in

liquid–liquid dispersions. Zhang et al (2008) suggested that most small rising bubble size
 125

distributions are well characterized by log-normal or gamma distributions in a gas–liquid

and gas–liquid–solid slurry bubble column reactor. On the other hand, Biswal et al (2008)

found that the bubble size distribution in the flotation follows the Rosin-Rommler equation.

Therefore, in the present study, normal, log-normal, Weibull (Rosin–Rammler) and gamma

distributions were fitted to the experimental data. The Normal (Gaussian) distribution is

characterized by a bell-shaped curve that is symmetrical around the mean; the log-normal

distribution is a probability distribution of a random variable whose logarithm is normally

distributed (Buzsáki and Mizuseki, 2014). Alternatively, the Weibull and gamma

distributions are described by the shape, scale, and threshold parameters (Silva and Lisboa

2007). The probability distribution functions of those four distributions have been

described in detail elsewhere (Fatima and Fortes, 1987; Silva and Lisboa 2007).

The Anderson-Darling (AD) test has been used to compare these distributions to identify

the best fit for the bubble size distribution in EF (Stephens, 1974). In this study, the

experimental data were processed for the AD test by statistical software MINITAB 15. In

an Anderson-Darling test, if the P-value is lower than 0.05 at the confidence interval of

95%, the data do not follow the specified distribution. The smallest Anderson-Darling

(AD) values identify the closest fit distribution.

4.3 Result and discussions

The AD values of different bubble size distributions of EF are presented in Table 4.2. As

seen the log-normal and gamma distributions are fairly similar according to the values of

AD. Silva and Lisboa (2007) also stated that the log normal and gamma distributions are

similar in shape for the same coefficient and are widely used for describing positively
 126

skewed data. The differences between the gamma and log-normal distribution become

most significant in their tail behavior (Henk, 2003, Silva and Lisboa 2007). The logarithm

of gamma distribution has more of a tail on the left, and less of a tail on the right. On the

other hand, the log-normal distribution is symmetric with a heavier right tail.

It is observed that the AD values for the log-normal and gamma distributions are close

(Table 4.2). However, a good argument in favor of the log-normal distribution is presented

based on the lower AD value. Moreover, since the P-values according to the AD log-

normality test for bubbles size distributions are consistently higher than 0.05, it can be

concluded that log-normality distribution is valid for bubble size distribution in the EF cell

(Alam et al, 2011). On the other hand, the P-values for the normal, and Weibull (Rosin–

Rammler) distributions are very low with high AD values. Therefore, bubbles size

distributions in EF do not fit well by either normal or Weibull distribution. The individual

probability plot for normal, log-normal, Weibull (Rosin–Rammler) and gamma

distributions also shows that in the log-normal distribution, the data points approximately

follow a straight line with the highest P-value. Example of this probability plot is shown in

Figure 4.3 based on the result of Exp. no. 10. Therefore, it can be concluded that the

experimental bubble size distributions were satisfactorily represented by the log-normal

distribution, in which the probability density function of bubble size (di) can be expressed

as (Fatima and Fortes, 1988):

1 (𝑙𝑛𝑑𝑖− 𝑑𝑏 )2
P (di) = 𝑑 𝜎√2𝜋 exp [- ] (4.9)
𝑖 2(𝜎)2

Where, σ and db are the standard deviation and arithmetic mean of lndi. Upon integration

of Eq. (4.9), the cumulative distribution function can be expressed as

 127

1 (𝑙𝑛𝑑𝑖− 𝑑𝑏 )
D (di) = 2[1+ erf [ ] (4.10)
𝜎√2

The experimental results of bubble size distribution will be discussed in the following

sections according to the simulated log-normal distributions (based on Eq. (4.9) and

(4.10)). In addition, the quality of data fitting to the log-normal function is shown

graphically in Figures 4.5 (a), 4.6(a), 4.8(a), 4.9(a), 4.12(a), 4.13(a), 4.15(a), 4.16(a),

4.18(a) and 4.20(a).

4.3.1 Effect of pH

To assess the effect of pH on bubble size of EF, a series of batch tests were performed (Exp

no 1-6). The Current density was kept constant throughout the series while the water pH

(2, 7 and 12) was varied. Figure 4.4 shows that the pH of the electrolyte medium can

significantly affect the mean bubble size of both hydrogen and oxygen bubbles. The

smaller mean bubble size of hydrogen (33.9 µm) was obtained under near neutral pH

values, whereas the mean bubble size of hydrogen in strongly acidic (pH =2) and alkaline

(pH = 12) media are larger, i.e. 65 and 50 µm, respectively. On the other hand, oxygen

formed on a Ti-IrO2 plate attained a mean bubble size of 32.7µm in an acid medium and

increased with pH increase. The trend is consistent with that of previous studies (e.g. Raju

and Khangaonkar, 1984; Glembotsky et al. 1973; Brandon and Kelsall, 1985). This trend

can be described by Eq. (4.8) which suggests that if the current density is constant (i.e. the

generating rate of O2 and H2 is constant) bubble dia is inversely proportional to the total

number of bubbles generate per unit time. Generally, a lower number of H2 bubbles are

observed at extremely low and high pH compared to that at neutral pH (Carlos, 2010).

Therefore, more nucleation sites become active on the cathode surface at neutral pH for
 128

which a decrease in H2 bubble diameter occurs at neutral pH. For similar reason, smaller

O2 bubbles are observed at low pH. It is also observed that when the pH is close to 6.0 the

mean bubble sizes of H2 and O2 come very close to each other (Figure 4.4) which is

consistent with the result of Raju and Khangaonkar (1984).

Figure 4.5 shows the simulated log-normal distribution of O2 bubble size at different pH.

As can be seen from the distribution curve, the amount of fine bubbles increased as the pH

decreased. Several researchers (e.g. Ketkar et al 1991) have stated that one of the

advantages of EF is to produce uniform bubble size. However Figure 4.5 shows that a wide

range of bubbles was produced with a difference of as much as 50 µm at pH 7 and pH 12.

The variations in H2 bubble size are illustrated in Figure 4.6. As seen in the figure, the finer

bubbles are generated at pH 7. On the other hand, at pH 2, H2 bubbles have a wider range.

Grau and Heiskanen (2005) stated that a broad range of bubble sizes might have a positive

effect on the flotation of coarse particles. In summary, this study reveals that the desired

bubble size and distribution can be manipulated by varying the pH value of the medium in

EF.

4.3.2 Effect of current density

The Current density is the most important parameter in an EF process. Effects of the current

densities on the bubble size and distributions are shown in Figures 4.7-4.9. The current

density in the experiments varied between 10 and 150 A/m2 while the other conditions

were kept constant. Bubbles generated at 10 A/m2 are comparatively large, and does not

show any sign of coalesce. On the other hand, at the current density 150 A/m2, the amount

of bubbles significantly increases, as described in the Faradays law, and the coalesce of
 129

bubbles is clearly observed (Figure 4.2). Figure 4.7 shows that the mean bubble size of

both O2 and H2 decreases with the increase in the current density up to 100 A/m2. It is well

established that the diameter of bubbles at the time of departure from the electrode

decreases as the current density is increased (Vogt, 1989). Because as current density

increases, more nucleation sites become active on the electrode surface and lead to generate

smaller bubbles (Lumanauw, 2000). However it is noteworthy to mention here that the

findings reviewed by Vogt (1989) or experiments conducted by Lumanauw (2000) strictly

refers to the diameter at detachment from an electrode; not for freely moving bubbles. In

the present study, a further increase in the current density above 100 A/m2 causes an

increase in the size of free moving bubbles. This is attributed to the fact that at higher

current densities, a large number of bubbles are produced, which facilitates the coalescence

process and form large bubbles.

Figure 4.8 shows the bubble size distribution of O2 for different current densities. As seen

in the figure, the increase in current density has a strong influence on the bubble size

distribution of O2. It was found that finer bubbles are produced at 100 A/m2. The variations

in H2 bubble size are illustrated in Figure 4.9. As seen, the bubble-size distribution was

shifted towards smaller bubble size as the current density was increased up to 100 A/m2

and again was shifted towards larger bubble size as the current density was increased to

150 A/m2.

As mention earlier that EF can control the number of bubbles by changing the current

density; the relationship between the number of bubbles and current density was

determined for Exp. nos. 2, 5 and 7-14. The amount of both oxygen and hydrogen bubbles

were counted under the same area (1 cm2) of different images for various current densities.
 130

The results are shown in Figure 4.10. The linear relationship between the current density

and the number of bubbles is in agreement with Eqs. (4.5) and (4.6). Moreover, it is

noteworthy to mention here that the number of hydrogen bubbles are almost twice than that

of oxygen bubbles under the same current density.

4.3.3 Effect of frother

Frothers are widely used in flotation to assist generation of small bubbles. The major

function of the frother is to conserve the size of bubble by preventing coalescence (Cho

and Laskowski, 2002). Moreover, recent studies (Kracht and Finch, 2009) have shown that

frothers also help to break down bubbles, thus reducing bubble sizes. Although the frother

plays an important role in froth flotation (Grau and Laskowski, 2006), the use of frother in

EF has not yet been extensively reported in open literature. To assess the effect of frother

(tennafroth 250) on bubble size of EF, a series of batch flotation tests were performed (Exp

nos. 25-30). The frother concentration in the experiments varied between 10 mg/L and 30

mg/L while the other conditions were kept constant. The measured bubble size vs frother

concentration is shown in Figure 4.11. As seen the bubble size decreased with increasing

frother dosage up to 20 mg/L, above which further increase of the frother concentration

does not affect the bubble size. Grau et al (2005) defined this point as the critical

coalescence concentration (CCC) of frother. Cho and Laskowski (2002) showed that the

coalescence is completely prohibited at the frother concentrations exceeding CCC values.

It is clear from Figure 4.11 that at frother dosages exceeding the CCC value of 20 mg/L for

both O2 and H2 bubbles, the conditions in the cell can be defined as non-coalescing. Figure

4.12 shows the O2 bubble size distributions measured under different frother

concentrations. As seen in the Figure (4.12), the amount of fine bubbles increased notably
 131

as the concentration of frother increased. The variations in H2 bubble size are illustrated in

Fig 4.13. It is found that the mean size of H2 decreased, and the number of finer bubbles

increased significantly as the concentration of frother increased.

4.3.4 Effect of Ionic strength

Dai et al. (2007) found that bubble size decreases with increasing the ionic strength of the

electrolyte solution. However Liuyi et al (2014) reported the opposite trend, which was

attributed to agglomeration of bubbles. Burns et al. (1997) suggested that there is no clear

trend of bubble diameters as a function of the ionic strength. Therefore, in the present study,

the effect of ionic strength (by adding KNO3) on the bubble size was investigated in a series

of experiments. The Current density and pH were kept constant throughout the tests while

the concentration of KNO3 was varied between 0.1 M and 0.5 M, under the same electrode

configuration. Figure 4.14 shows a slight decrease in the bubble size with increasing ionic

strength, which supports the findings of Dai et al (2007). The findings might be explained

by two previous studies conducted by Mÿmicci and Nicoderno (1967) and Ziemenski and

Whittemore (1971). Ziemenski and Whittemore (1971) found that adding salt in solution

increase the viscosity of the electrolyte and hence increase the rigidity of the surface films

of the bubbles so that the coalescence is more difficult. Furthermore, Mÿmicci, and

Nicoderno (1967) suggested that adding salt would increase the electric repulsive forces

between bubbles for their high surface potential, which would also help to avoid collision.

Figures 4.15 and 4.16 show the bubble size distribution of O2 and H2 respectively for

different KNO3 concentrations. It was found that the bubble-size distribution was shifted

slightly towards the smaller sizes as the concentration of KNO3 was increased for both O2
 132

and H2. Therefore, it can be concluded that KNO3 would reduce bubble sizes, however it

is to less extent comparing to the frother.

4.3.5 Effect of electrode materials and geometry

The mean H2 bubble sizes as a function of current density (from 10 to 150 A/m2) for the

SS plate and SS mesh electrode are shown in Figure 4.17. The SS mesh electrodes produce

the largest bubble (101.5 μm) diameter within the current density ranges studied. Bubbles

generated from the SS mesh might have a larger contact angle and consequently produce

longer bubble foot perimeter that holds the bigger bubble on the surface (Lumanauw,

2000). Figure 4.18 shows the H2 bubble size distributions for the SS plate and SS mesh

electrode. It is found that SS plate produces more finer bubbles with compared to that of

SS mesh electrodes.

The mean bubble measurements as a function of current density (10 to 150 A/m2) for the

SS 316 plate and Ti/IrO2 plate electrode are shown in Figure 4.19. As seen in the figure,

the Ti/IrO2 plate produces comparatively smaller O2 bubbles under the same operating

condition. The difference in wettability (related to the adhesion force) of the two materials

may have impact on the bubble size (Ibl, and Venczel, 1970). Figure 4.20 compares the

bubble size distributions for the SS 316 plate and Ti/IrO2 plate electrode. It is found that

the Ti/IrO2 plate produces finer bubbles compared to that of the SS plate.
 133

4.4 Summary

The following conclusions are reached based on the present study:

 The experimental data obtained from the tests conducted in the lab scale EF cells

were well represented by the log-normal distribution.

 The water pH has significant effect on bubble sizes. The smaller bubbles of H2 at

the cathode were observed at near neutral pH values, whereas smaller bubbles of

O2 at the anode were observed in low pH water.

 The Current density has significant effect on the bubble size. The mean bubble size

decreases with the increase in the current density up to a threshold value of 100

A/m2. Further increase in the current density causes an increase in bubble sizes,

which is attributed to the coalescence of free moving bubbles.

 The study on the frother (tennafroth 250) shows that with increasing frother

concentration, the bubble size decreased until a critical coalescence concentration

(CCC) was reached. At concentrations above CCC, no further changes in the bubble

size were observed.

 A slight decrease in the bubble size was observed with the increase of ionic strength

(by adding KNO3).

 It is found that when used as the anode, the Ti/IrO2 plate produces finer O2 bubbles

than the stainless steel SS 316 plate.

 134

4.5 References

Ahmed, N., Jameson, G.J. 1985. The effect of bubble size on the rate of flotation of fine
particles. International Journal of Mineral Processing 14, 195–215.

AI-Hayes, R. A. M., and Winterton, R. H. S. 1981. Int. J. Heat Mass transfer, vol. 24,
p.213-331.

Alam, R., Shang, J.Q. and Cheng, X. 2011. Optimization of digestion parameters for
analyzing the total sulphur of mine tailings by inductive coupled plasma optical
emission spectrometry. Environmental Monitoring and Assessment (Springer), Vol.
184, No. 5, pp. 3373-3387.

Bande, R.M., Prasad, B., Mishra, I.M., Wasewar, K.L. 2008. Oil field effluent water
treatment for safe disposal by electro-flotation. Chem. Eng. J., 137: 503–509.

Beer, H. B. 1972. U.S. Patent 3,632,498.

Bennet, A. J. R., Chapmen, W. R., Dell, C. C. 1958. Studies in the froth flotation of coal
Third International Coal Preparation Conngress, Brussels-Leige.

Ben-Yosef, N.; O. Ginio, D. Mahlab, and A. Weitz, 1974. Bubble size distribution
measurement by Doppler velocimeter. Journal of Applied Physics 46, 738 (1975);
doi: 10.1063/1.321638

Biswal, S.K., Reddy, P. S. R. and Bhaumik S. K. 2009. Bubble size distribution in a

flotation column, The Canadian Journal of Chemical Engineering ; 72(1):148 - 152.
DOI: 10.1002/cjce.5450720123

Brandon, N. P., Kelsall, G. H. 1985. Interfacial electrical properties of electrogenerated

bubbles. Journal of Applied Electrochemistry, 15, 485-493.

Burns, S. E., Yiacoumi, S., Tsouris, C. 1997. Microbubble generation for environmental
and industrial separations. Separation and purification technology, 11, 221-232.

Buzsáki G. & Mizuseki K. 2014. Box 1: Normal and log-normal distributions, Nature
Reviews Neuroscience 15, 264–278

Carlos Jim´enez, Beatriz Talavera, Cristina S´ aez, Pablo Canizares and Manuel A.
Rodrigo. 2010. Study of the production of hydrogen bubbles at low current densities
for electro-flotation processes. J Chem Technol Biotechnol; 85: 1368–1373

Chen, X., Chen, G. 2010. Electro-flotation in C. Comninellis and G. Chen (eds.),

Electrochemistry for the Environment, Springer Science+Business Media, LLC. 263-
77.
 135

Cho, Y.S., Laskowski, J.S., 2002. Effect of flotation frothers on bubble size and foam
stability. International Journal of Mineral Processing 64, 69–80.

Dai J, Xie G, Liu S, Wang X. 2007. Analysis of influencing factors of flotation bubble size.
Coal Prep Technol; 3:7–10.

Fatima, M. and Fortes, M.A. 1988. Grain Size Distribution: The log-normal and the gamma
distribution function, Scripta Metallurgica, vol 22, 35-40.

Fukui, Y., Yuu, S. 1985. Removal of colliodal particles in electro-flotation. AIChE Journal,
31(2), 201-208.

Glembotsky, V. A., Mamakov, A. A., Sorokina, V. N. 1973. Electroannaya Obrabotka

Materialov, 5, 66.

Grau, R.A. K. Heiskanen. 2005. Bubble size distribution in laboratory scale flotation cells
Minerals Engineering 18. 1164–1172

Grau, R.A., Laskowski, J.S., 2006. Role of frothers in bubble generation and coalescence
in a mechanical flotation cell. The Canadian Journal of Chemical Engineering 84,
170–182.

Grau, R.A., T, Janusz S. Laskowski, Kari Heiskanen (2005) Effect of frothers on bubble
size Int. J. Miner. Process. 76 225– 233

Han MY, Kim MK and Ahn HJ. 2006. Effects of surface charge, microbubble size and
particle size on removal efficiency of electro-flotation. Water Sci Technol 53:127–
132.

Heiskanen, K., 2000. On the relationship between flotation rate and bubble surface area
flux. Minerals Engineering 13 (2), 141–149.

Henk C. Tijms, 2003. A First Course in Stochastic Models. Vrije Universiteit, Ámsterdam,
Wiley, the Netherlands, 2003

Ibl, N. and Venczel, I. 1970. Metalloberfiache, vol. 34. p. 365.

Janssen, L. J. J., Hoogland, J. G. 1973. Electrochim Acta, 18, 543.

Ketkar, D.R., Mallikarjunan, R., Venkatachalam, S.1991. Electro-flotation ofquartz fines.

Int J Mineral Process 31:127–138.

Khosla, N. K., Venkatachalam, S., Somasundaran, P. 1991. Pulsed electrogeneration of

bubbles for electro-flotation. Journal of applied electrochemistry. Vol. 21, pp986-
990.

Kracht, W., J.A. Finch. 2009. Bubble break-up and the role of frother and salt. Int. J. Miner.
Process. 92 (2009) 153–161
 136

Landolt, D., Acosta, R., Muller, R. H., Tobais, C. W. 1970. An optical study of cathodic
hydrogen evolution in high rate electrolysis. Journal of the Electrochemical Society,
117(6), 839-845.

Lee, S. M. 1969. Effect of equivalent bubble sizes on the flotability of single bubble forth
flotation. Journal of AIChE, 7, 202-213.

Liuyi, R., Zhang Y., Qin W., Bao S., Wang P., Yang C. 2014. Investigation of condition-
induced bubble size and distribution in electro-flotation using a high-speed camera
International Journal of Mining Science and Technology 24. 7–12

Lumanauw, D. 2000. Hydrogen bubble characterization in alkaline water electrolysis,

thesis of Master of Applied Science, Department of Metailurgy and Materiais
Science, University of Toronto

Mÿmicci, G. ÿnd Nicoderno, L. 1967. Chern. Eng. Sci. voi. 27, p. 1257.

Pacek, A., Man, C., Nienow, A., 1998. On the Sauter mean diameter and size distributions
in turbulent liquid/liquid dispersions in a stirred vessel. Chemical Engineering
Science 53 (11), 2005–2011.

Raju G. B., Khangaonkar.1984. Electro-flotation–A critical review, Trans. Indian Inst.

Metals 37(1), 59–66.

Sarkar, M.S.K.A., .Evans, G. M., Donne, S. W. 2010. Bubble size measurement in electro-
flotation. Minerals Engineering 23, 1058–1065

Sarkar, S.K.A .2012. Electroflotation: its application to water treatment and mineral
processing. PhD thesis, chemical engineering dept. The University of Newcastle.

Sides, P. J. 1986. Phenommenon and effects of electrolytic gas evolituion. Modern Aspects
of Electrochemistry, R. E. White, J. O. N. Bokris, and B. E. Conway, eds., Plenum
Press, New York, NY, 303-54.

Silva, EL and Lisboa, P. 2007. Analysis of the characteristic features of the density
functions for gamma, Weibull and log-normal distributions through RBF network
pruning with QLP Proceedings of the 6th WSEAS Int. Conf. on Artificial
Intelligence, Knowledge Engineering and Data Bases, Corfu Island, Greece,
February 16-19, 2007

Stephens, M. A. 1974. EDF Statistics for Goodness of Fit and Some Comparisons, Journal
of the American Statistical Association, 69, pp. 730-737.

Tavlarides, L., Stamatoudis, M., 1981. The analysis of interphase reactions and mass
transfer in liquid–liquid dispersions. Advances in Chemical Engineering 11, 199–
273.
 137

Vogt H. 1989 The Problem of the Departure Diameter of Bubbles at Gas-Evolving

Electrodes, Electrochemica Acta, Vol 34, No 10, pp 1429-1432

Vogt, H. 1983. In Comprehensive treatise of electrochemistry (edited by E. Yeager, J.

O’M. Bockris, B. E. Conway and S. Sarangapani), Plenum Press, New York, Vol. 6,
445.

Yoon, R.-H. 2000. Int. J. Miner. Process. 58,129–143

Zhang, L., Tao Li, Wei-yong Ying, Ding-ye Fang. 2008. Rising and descending bubble
size distributions in gas–liquid and gas–liquid–solid slurry bubble column
reactor.chemical engineering research and design 8 6 1143–1154

Ziemenski, S. A. and Whittemore, R. C. 1971. Chem. Eng. Sci., vol. 26, p. 509-520.
 138

TABLES

Table 4.1 Experimental Condition

Experiment pH Current Electrode Frother Concentration Gas

No Density configuration concentration of KNO3 (M)

(A/m2) (Anode/Cathode) (mg/L)

1 2 100 TP/SP 0 0.1 O2

2 7 100 TP/SP 0 0.1 O2

3 12 100 TP/SP 0 0.1 O2

4 2 100 SP/SP 0 0.1 H2

5 7 100 SP/SP 0 0.1 H2

6 12 100 SP/SP 0 0.1 H2

7 7 10 TP/SP 0 0.1 O2

8 7 25 TP/SP 0 0.1 O2

9 7 50 TP/SP 0 0.1 O2

10 7 150 TP/SP 0 0.1 O2

11 7 10 SP/SP 0 0.1 H2

12 7 25 SP/SP 0 0.1 H2

13 7 50 SP/SP 0 0.1 H2

14 7 150 SP/SP 0 0.1 H2

15 7 10 SP/SP 0 0.1 O2

16 7 25 SP/SP 0 0.1 O2

17 7 50 SP/SP 0 0.1 O2
 139

18 7 100 SP/SP 0 0.1 O2

19 7 150 SP/SP 0 0.1 O2

20 7 10 SM/SM 0 0.1 H2

21 7 25 SM/SM 0 0.1 H2

22 7 50 SM/SM 0 0.1 H2

23 7 100 SM/SM 0 0.1 H2

24 7 150 SM/SM 0 0.1 H2

25 7 150 TP/SP 10 0.1 O2

26 7 150 TP/SP 20 0.1 O2

27 7 150 TP/SP 30 0.1 O2

28 7 150 SP/SP 10 0.1 H2

29 7 150 SP/SP 20 0.1 H2

30 7 150 SP/SP 30 0.1 H2

31 7 150 TP/SP 0 0.25 O2

32 7 150 TP/SP 0 0.5 O2

33 7 150 SP/SP 0 0.25 H2

34 7 150 SP/SP 0 0.5 H2

 140

Table 4.2 AD values and corresponding P-values for different distributions

Experiment Normal Log-normal Weibull Gamma Best fitted

No distribution distribution
(Rosin– based on
Rammler) lowest AD-
value

1 0.453 0.222 0.600 (P = 0.247 (P > log-normal

(P=0.258) (P=0.817) 0.117) 0.250)

2 1.205 0.430 1.837 (P < 0.611 (P = log-normal

(P<0.005) (P=0.300) 0.01) 0.119)

3 0.638 0.618 0.654 (P 0.624 (P = log-normal

(P=0.09) (P=0.101) =0.085) 0.0.107)

4 0.393 0.306 0.519 (P = 0.342 (P > log-normal

(P=0.347) (P=0.538) 189) 0.250)

5 0.513 0.342 1.176 0.312 (P > gamma

(P=0.188) (P=0.484) (P<0.01) 0.250)

6 0.292 0.207(P=0.860) 0.443 (P > 0.209 (P > log-normal

(P=0.590) 0.250) 0.250)

7 0.526 0.478 0.697 (P = 0.469 (P > gamma

(P=0.155) (P=0.207) 0.061) 0.250)

8 1.589 (P< 0.656 2.061 (P < 0.897 (P = log-normal

0.005) (P=0.083) 0.010) 0.023)

9 1.521 (P< 0.489 1.973 (P < 0.730 (P = log-normal

0.005) (P=0.214) 0.010) 0.060)

10 1.070 0.280 1.278 (P < 0.460 (P> log-normal

(P=0.008) (P=0.635) 0.010) 0.250)
 141

11 0.363(P= 0.322 (P= 0.500 (P 0.321 (P gamma

0.405) 0.506) =0.205) >0.250)

12 0.469 0.176 0.710 (P = 0.224 (P log-normal

(P=0.240) (P=0.919) 0.062) >0.250)

13 0.287 0.217 0.675 (P = 0.180 (P gamma

(P=0.613) (P=0.837) 0.078) >0.250)

14 1.057 0.441 1.476 (P 0.608 (P = log-normal

(P=0.008) (P=0.280) <0.01) 0.121)

15 0.666 0.383 1.042 (P < 0.477 (P = log-normal

(P=0.076) (P=0.381) 0.010) 0.243)

16 0.653 0.416 1.008 (P < 0.495 (P = log-normal

(P=0.082) (P=0.318) 0.010) 0.227)

17 0.608 0.473 0.727 (P = 0.505 (P = log-normal

(P=0.108) (P=0.233) 0.054) 0.218)

18 0.248 0.278 0.338 0.239 (P = gamma

(P=0.736) (P=0.635) (P>0.250) >0.250)

19 0.968 0.362 1.55 (P < 0.479 (P = log-normal

(P=0.013) (P=0.427) 0.010) 0.242)

20 0.356 0.359 0.584 (P = 0.296 (P > gamma

(P=0.447) (P=0.439) 0.135) 0.250)

21 0.398 0.144 0.506 (P = 0.194 (P > log-normal

(P=0.349) (P=0.967) 0.205) 0.250)

22 1.797 0.566 1.264 (P < 0.884 (P = log-normal

(P<0.005) (P=0.136) 0.010) 0.025)

23 0.557 0.155 0.956 (P = 0.223 (P > log-normal

(P=0.145) (P=0.954) 0.016) 0.250)

24 1.072 0.239 2.302 (P < 0.346 (P > log-normal

(P=0.008) (P=0.773) 0.010) 0.250)
 142

25 0.443 (P= 0.396 (P= 0.783 (P = 0.400 (P > log-normal

0.270) 0.352) 0.039) 0.250)

26 1.077 0.743 1.733 (P < 0.817 (P = log-normal

(P=0.008) (P=0.051) 0.010) 0.037)

27 0.638 0.254 1.184 (P < 0.329 (P > log-normal

(P=0.090) (P=0.714) 0.010) 0.250)

28 0.248 (P= 0.278 0.338 (P > 0.239 (P > gamma

0.736) (P=0.635) 0.250) 0.250)

29 0.441 0.476 0.737 (P = 0.423 (P > gamma

(P=0.275) (P=0.226) 0.049) 0.250)

30 0.775 0.696 1.618 (P 0.664 (P = gamma

(P=0.042) (P=0.067) <0.010) 0.086)

31 0.356 0.242 (P= 0.552 0.188 (P> log-normal

(P=0.449) 0.762) (P=0.166) 0.250

32 0.233 0.152 (P= 0.519 (P= 0.111 (P> log-normal

(P=0.791) 0.958) 0.197) 0.250)

33 0.725 0.203 (P= 1.207 0.316 (P> log-normal

(P=0.056) 0.873) (P<0.010) 0.250)

34 0.467 (P= 0.206 (P= 0.813 (P= 0.218 (P> log-normal

0.243) 0.864) 0.034) 0.250)
 143

FIGURES

Figure 4.1 Schematic diagram of the Experimental set-up

 144

Bubble
coalescence at
higher current
density

Bigger bubble
forms from
coalescence of
two bubbles

Figure 4.2 Bubble coalescence at high current density

 145

Exp 10
N ormal - 95% C I Lognormal - 95% C I G oodness of F it Test
99.9 99.9
N ormal
99 99 A D = 1.070
P -V alue = 0.008
90 90
P er cent

P er cent
Lognormal
50 50 A D = 0.280
P -V alue = 0.635
10 10
Weibull
1 1 A D = 1.278
0.1 0.1 P -V alue < 0.010
0 50 100 20 50 100
G amma
Weibull - 95% C I G amma - 95% C I A D = 0.460
99.9 99.9 P -V alue > 0.250

90 99

50 90
P er cent

P er cent
50
10

10
1
1

0.1 0.1
10 100 20 50 100

Figure 4.3 probability plot for different types of distribution

 146

H2 bubble O2 bubble

70
65

Mean bubble dia (μm)

60
55
50
45
40
35 pH at which the mean size of H2 and O2 are
30
approximately equal
25
20
0 2 4 6 8 10 12 14
pH

Figure 4.4 Effect of pH on bubble size

 147

O2 bubble
pH 2
100

Cumulative Bubble size distributions (%)

pH 7
pH 12

80

60

40

20

0
30 45 60 75 90
Bubble dia (µm)

(a)

O2 bubble
pH 2
25 pH 7
pH 12

20
Number of bubbles (%)

15

10

0
30 45 60 75 90
Bubble dia (µm)

(b)

Figure 4.5: Bubble size distribution of O2 measured at different pH: (a)

comparison of experimental data with the simulated cumulative log-normal
distribution (b) Simulated log-normal distribution of bubbles
 148

H2 bubble
pH 2
100

Cumulative Bubble size distributions (%)

pH 7
pH 12

80

60

40

20

0
30 45 60 75 90
Bubble dia (µm)

(a)

H2 bubble
pH 2
30 pH 7
pH 12

25
Number of bubbles (%)

20

15

10

0
30 45 60 75 90
Bubble dia (µm)

(b)

Figure 4.6: Bubble size distribution of H2 measured at different pH: (a)

comparison of experimental data with the simulated cumulative log-normal
distribution (b) Simulated log-normal distribution of bubbles
 149

O2 bubbles H2 bubble

65
60
Mean bubble dia (μm)
55
50
45
40
35
30
25
20
0 20 40 60 80 100 120 140 160
Current Density, A/m2

Figure 4.7 Effect of current density on mean bubble size

 150

O2 bubble
10 A /m2
100

Cumulative Bubble size distributions (%)

25 A /m2
50 A /m2
100 A /m2
150 A /m2
80

60

40

20

0
30 40 50 60 70 80 90
Bubble dia (µm)

(a)

O2 bubble
10 A /m2
25 A /m2
20 50 A /m2
100 A /m2
150 A /m2
Number of bubbles (%)

15

10

0
30 40 50 60 70 80 90
Bubble dia (µm)

(b)

Figure 4.8: Bubble size distribution of O2 measured at different Current

density: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 151

H2 bubble
10 A /m2
100

Cumulative Bubble size distributions (%)

25 A /m2
50 A /m2
100 A /m2
150 A /m2
80

60

40

20

0
20 30 40 50 60 70 80
Bubble dia (µm)

(a)

H2 bubble
10 A /2
30 25 A /m2
50 A /m2
100 A /m2
25 150 A /m2
Number of bubbles (%)

20

15

10

0
20 30 40 50 60 70 80
Bubble dia (µm)

(b)

Figure 4.9: Bubble size distribution of H2 measured at different Current

density: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 152

No of bubbles/ cm2 Oxygen generation rate (L/sec) by Faraday's law

120 0.00014

Oxygen generation rate (L/sec)

100 0.00012

Number of O2 0.0001
80
bubbles 0.00008
60
0.00006
40
0.00004
20 0.00002
0 0
0 50 100 150 200
Current Density, A/m2

(a)

No of bubbles/cm2 Hydrogen generation rate (L/sec) by Faraday's law

250 0.0003

Hydrogen generation rate (L/sec)

0.00025
Number of H2 bubbles

200
0.0002
150
0.00015
100
0.0001
50 0.00005

0 0
0 20 40 60 80 100 120 140 160
Current Density, A/m2

(b)

Figure 4.10 Effect of current density on number of bubbles/cm2: (a) O2 bubbles (b) H2
bubbles
 153

O2 bubbles H2 bubble

55

Mean bubble dia 50

45
40
(μm)

35
30
25 CC
20 C
0 5 10 15 20 25 30 35
Frother concentration (mg/L)

Figure 4.11 Effect of Frother (tennafroth 250) concentration on mean

bubble size
 154

O2 bubble
No frother added
100

Cumulative Bubble size distributions (%)

10 mg/L
20 mg/L
30 mg/L

80

60

40

20

0
30 40 50 60 70 80 90
Bubble dia (µm)

(a)

O2 bubble
35 No frother added
10 mg/L
20 mg/L
30 30 mg/L
Number of bubbles (%)

25

20

15

10

0
30 40 50 60 70 80 90
Bubble dia (µm)

(b)

Figure 4.12: Bubble size distribution of O2 measured at different frother

concentration: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 155

H2 bubble
No frother added
100

Cumulative Bubble size distributions (%)

10 mg/L
20 mg/L
30 mg/L

80

60

40

20

0
30 40 50 60 70
Bubble dia (µm)

(a)

H2 bubble

35 No frother added
10 mg/L
20 mg/L
30 30 mg/L
Number of bubbles (%)

25

20

15

10

0
30 40 50 60 70
Bubble dia (µm)

(b)

Figure 4.13: Bubble size distribution of H2 measured at different frother

concentration: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 156

O2 bubble H2 bubble

55

Mean dia of bubbles (μm)

50
45
40
35
30
25
20
0 0.1 0.2 0.3 0.4 0.5 0.6
Concentration KNO3 (M)

Figure 4.14 Effect of KNO3 concentration on mean bubble size

 157

O2 bubble
0.1 M
100

Cumulative Bubble size distributions (%)

0.25 M
0.5 M

80

60

40

20

0
20 30 40 50 60 70 80 90
Bubble dia (µm)

(a)

O2 bubble
0.1 M
20 0.25 M
0.5 M
Number of bubbles (%)

15

10

0
20 30 40 50 60 70 80 90
Bubble dia (µm)

(b)

Figure 4.15: Bubble size distribution of O2 measured at different KNO3

concentration: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 158

H2 bubble
0.1 M
100

Cumulative Bubble size distributions (%)

0.25 M
0.5 M

80

60

40

20

0
30 40 50 60 70
Bubble dia (µm)

(a)

H2 bubble
25 0.1 M
0.25 M
0.5 M

20
Number of bubbles (%)

15

10

0
30 40 50 60 70
Bubble dia (µm)

(b)

Figure 4.16: Bubble size distribution of H2 measured at different KNO3

concentration: (a) comparison of experimental data with the simulated cumulative
log-normal distribution (b) Simulated log-normal distribution of bubbles
 159

SS plate SS mesh

80

70

Mean bubble dia (μm) 60

H2

50

40

30

20
0 20 40 60 80 100 120 140 160
Current Density, A/m2

Figure 4.17 Effect of geometry of Cathode on mean bubble size

 160

H2 bubble
SS plate
100

Cumulative Bubble size distributions (%)

SS mesh

80

60

40

20

0
24 32 40 48 56
Bubble dia (µm)

(a)

H2 bubble

25 SS Plate
SS mesh

20
Number of bubbles (%)

15

10

0
24 32 40 48 56
Bubble dia (µm)

(b)

Figure 4.18: Bubble size distribution of H2 measured for SS plate and Mesh: (a)
comparison of experimental data with the simulated cumulative log-normal
distribution (b) Simulated log-normal distribution of bubbles
 161

SS 316 Ti/IrO2

65
60
mean bubble dia (μm) 55
50
45
O2

40
35
30
25
20
0 20 40 60 80 100 120 140 160
Current Density A/m2

Figure 4.19 Effect of Anode materials on mean bubble size

 162

O2 bubble
Ti/IrO2 plate
100
Cumulative Bubble size distributions (%)
SS 316 plate

80

60

40

20

0
30 40 50 60 70 80
Bubble dia (µm)

(a)

O2 bubble
Ti/IrO2 plate
SS 316 Plate
20
Number of bubbles (%)

15

10

0
30 40 50 60 70 80
Bubble dia (µm)

(b)

Figure 4.20: Bubble size distribution of O2 measured for Ti/IrO2 and SS 316 plate:
(a) comparison of experimental data with the simulated cumulative log-normal
distribution (b) Simulated log-normal distribution of bubbles
 163

CHAPTER 5

REMOVAL OF BITUMEN FROM MATURE OIL SANDS TAILINGS SLURRIES

BY ELECTRO-FLOTATION

5 Introduction

The oil sands deposit in northern Alberta, Canada, is estimated to contain 174 billion

barrels of bitumen (Alberta Energy and Utilities Board 2005). The Canadian oil sands

industries withdraw an average of 3 barrels of freshwater for every barrel of oil produced

(Allen, 2008). The resulting process water of the oil sands mining operation is discharged

in large tailings ponds. Allen (2008) has identified some target pollutants in the tailings

ponds’ water, which should be reduced to an allowable level so that the tailings can be used

directly in the reclamation process (Elliott et al 2010). Residual bitumen, one of the target

pollutants poses a hazard to aquatic biota, and its biodegradation could be a source of

Naphthenic Acid (Quagraine et al. 2005). Bitumen would disturb aquatic communities in

reclamation ponds, and contribute to the fouling of advanced treatment technologies.

Bitumen concentrations (measured as oil and grease) in tailings ponds water have been at

least 2.5 to 9 fold higher than the EPEA maximum discharge limit of 10 mg/L, thus up to

90% removal may be required (Allen, 2008).

Flotation has been recognized as the most economical way to separate bitumen since it has

a similar density to water (Masliyah et al 2004; Bichard, 1987). During extraction of

bitumen, caustic or sodium hydroxide is added to increase oil sands slurry pH, making

solids and bitumen more negatively charged, thereby helping bitumen liberation from sand
 164

grains (Lau et al 2013, Zhou et al, 2010, Dai, and Chung, 1995). Therefore, the pH of MFT

slurries often ranges from 8 to 9, which makes the bitumen surface fairly hydrophobic with

contact angles greater than 70 degrees (Zhou, 2002). In addition, the caustic environment

of the slurry induces ionization of natural surfactants (carboxylates and

sulphonates/sulphates) for which no frother is needed in bitumen flotation for stabilizing

bubbles (Zhou et al, 2010). Therefore, flotation can be considered as the most viable

technology to remove bitumen from tailings pond.

For flotation of oil, fine gas bubbles are introduced into the slurry. The bubbles attach to

the bitumen and the buoyancy lifts bitumen to the surface, which is then collected by

skimming. Gas bubbles can be produced through many methods, for example, dissolved

air-flotation, electrostatic spraying of air, and electrolysis, which is the principle of electro-

flotation (Burns, 1997). Conventional air flotation generates large bubbles, ranging in sizes

from 600 to 2000 μm (Wills, 1997). Some spargers can produce medium bubble sizes from

100 to 600 μm (Montes-Atenas et al, 2010). Guoxing et al (2004) reported from an

experimental study that bubbles sizes lower than 120 μm are favorable for flotation of

bitumen. Electro flotation (EF) is capable of producing hydrogen and oxygen bubbles

consistently smaller than 120 μm from water electrolysis. In EF, an electrolyte solution is

brought between two electrodes, and a direct current (dc) is passing through from the

positive pole (anode) to the negative pole (cathode). As a result of electrolysis reactions,

hydrogen and oxygen are formed at cathode and anode, respectively, in the form of gas

bubbles. The redox reactions at the electrodes can be expressed as (Chen and Chen, 2010):
 165

At the anode,

2H2O → O2 + 4H++ 4e (5.1)

At the cathode,

4H++ 4e → 2H2 (5.2)

The EF technique is versatile and competitive with other flotation techniques, such as

dissolved and dispersed air flotation (Burns et al, 1997). EF units are small, compact and

require lower maintenance and running costs than other flotation units (Kolesnikov, 1994).

In general, EF has three principle features that are advantageous comparing to other

flotation techniques, (Bande et al, 2008): i.e. 1. Extremely fine dispersed gas bubbles those

are uniform in sizes. This increases the surface area of contact between oil drops and gas

bubbles. 2. A specific gas bubble concentration and size can be generated by controlling

the current density, thereby increasing the probabilities of bubble and oil drop collisions.

3. The process can be optimized by the selection of an appropriate electrode surface and

solution conditions. In addition, it can achieve separation in much shorter retention time

compared to that of air flotation (Ibrahim et al, 2001). Applications of EF are especially

attractive for separation of oil from oil-water emulsions because of the low density of oil,

which has been reported extensively in the literature (Matis and Peleka, 2010, Hosny, 1996,

Mansour and Chalbi, 2006, Ibrahim et al. 2001, Vasibenko et al, 1973; Belyacva et al,

1980; Balmer and Foulds 1986; Il’in 2002; Il’in and Sedashova 1999). EF is also

successfully implemented for separation of paint solids from automotive assembly paint

booths without using any chemical agents or compressed air (Xu and Shang, 2009). EF is
 166

accepted as a cost effective means of treating effluents in Europe, where they compete

strongly with established process such as dissolved air flotation (Sarkar, 2012).

The objectives of this study are 1) to study the effect of EF for removal of bitumen from

oil sand tailings slurry; the target is to reduce the bitumen concentration to below the EPEA

maximum discharge limit (10 mg/L); and 2) to investigate the EF performance related to

variations of operating parameters (Current density, pH, bitumen concentration and NaCl

dosage).

5.1 Experimental

5.1.1 Mature fine tailings slurry

The flotation experiments were carried out using MFT samples obtained from a tailings

disposal pond in Fort McMurray, Alberta, Canada. The properties of Mature Fine Tailings

(MFT) have been studied (e.g. Alam et al, 2014; Guo, 2012), with a summary presented in

Table 5.1. The tailings can be characterized as MH according to the Unified Soil

Classification system (ASTM D2487), with about 80% silt-sized grains, and 20% clay

sized grains, 54.42% liquid limit, 36.04% plastic limit and plasticity index of 18.38.

Mineralogical analysis using X-ray diffractometry revealed that the mature fine oil sands

tailings contain 26-30 % quartz (SiO2), 40-50% illite (KAl2SiO3AlO10(OH)2), 11-25%

kaolinite (Al2Si2O5(OH)4) and trace amount of goethite (FeO(OH)) and biotite

(K(Mg,Fe)3(AlSi3O10)(F,OH)2) (Guo, 2012). The water content of the tailings as received

was found in the range of 158 % to 171%. The specific gravity of the sample was between

2.5-2.6. The dry density and void ratio of the sample were 0.49 g/cm3 and 4.26 respectively.

The organic matter of the sample was found in the range of 14.5 to 18%, of which a
 167

significant portion is bitumen. The pH of the tailings sample was 8.5, which makes the

slurry favorable to flotation of bitumen due to its effects on the physical properties of the

system, such as interfacial tension (Drelich, 2008) and electric surface potential (Czarnecki

et al, 2005).

Oil sands slurry was synthesized by mixing MFT with hot distilled water (80ºC) for 3

hours. The slurry was further diluted before the EF treatment by mixing with distilled water

for another 3 hours to achieve the desired concentrations of bitumen in the slurry to

replicate the actual tailings pond water.

5.1.2 Flotation

The laboratory scale studies were carried out in a flotation cell, as shown in Figure 5.1. The

cell has a volume of 4.5 L, with the length 22 cm at the top and 15.4 cm at the bottom, the

width 16.7 cm, and the height 25.2 cm. The cathode was placed horizontally at the top of

the anode, which is made of a stainless steel mesh (SS 316), an Iridium dioxide coated

titanium (Ti) mesh was used the anode. The spacing between the electrodes was 15 mm.

This electrode configuration has shown to be effective for quick dispersion of small gas

bubbles (Xu and Shang, 2009). The dc current for the circuit was maintained by a DC

power supply (HP 6545A). The experiments were carried out under various operating

conditions, including current density (50 A/m2 to 300 A/m2), bitumen concentration (22.48

mg/L to 314.72 mg/L), pH (3.3 to 10.68) and NaCl dosage (17.5 g/kg to 70 g/kg) (for

adjustment of water electrical conductivity). A summary of the flotation test conditions is

presented in Table 5.2. The pH was adjusted to a desired value through HCl or NaOH

solutions just before the test. After conditioning the slurry, the DC power was switched on

and regulated to a selected current density. The experiments were carried out by stirring
 168

the oil sand slurry at an agitation speed of 175 rpm for the duration of 90 min, during this

period the froth was collected. Samples of 10 ml were taken from the supernatant at a depth

of 50 mm below the free surface of the flotation cell at 10, 20, 30, 60 and 90 min during

each test. The bitumen removal rate, R (%) from the flotation unit was evaluated by using

the following equation:

𝐶
R (%) = (1- 𝐶 ) 100 (5.3)
0

Where, C0 and C are the initial and final concentrations of bitumen (mg/L) respectively.

The concentration of bitumen (as oil and grease) was determined with a FTIR spectrometer

(Nicolet 6700) using the wave number of 2920 cm-1 by ASTM D7575 (See appendix B).

The oil and grease extractor and the standards were supplied by Orono Spectral Solutions

Inc.

5.2 Results and discussions

5.2.1 Flotation Kinetics

Typically the effectiveness of flotation separation is measured by the percentage removal

of target pollutant; however the flotation kinetics represents a positive and more accurate

understanding for the process (Schuhmann, 1942). The Flotation kinetics studies the

deviation of concentration of target species as related to flotation time. Most authors (Alam

and Shang, 2012; Hernainz and Calero, 1996) compare the flotation process as analogous

to chemical kinetics. The majority of researchers have proposed the first order flotation

kinetics for EF of oil (e.g. Hosny, 1996, Mansour and Chalbi, 2006). A generalized

expression analogous to chemical kinetics can be used to represent the flotation process:
 169

𝑑𝐶
= -KCn (5.4)
𝑑𝑡

Where, C is the concentration of bitumen at time t, K is flotation rate constant (min-1) and

n is the kinetic order. In this study, the first (n = 1) order rate equations have been used to

determine flotation rate constant.

In the First Order Rate Equation:

𝑑𝐶
= -KC (5.5)
𝑑𝑡

Upon integration Eq. 5.5 becomes

𝐶0
ln = Kt (5.6)
𝐶

Eq. (5.6) can be rewritten as

C = C0𝑒 −K𝑡 (5.7)

Where, C0 is the initial concentration of the bitumen in MFT slurries. According to

𝐶0
Equation 5.6, ln and flotation time, t, yield a linear relationship if K is constant (Alam
𝐶

et al, 2012). An Example of this relationship is shown in Figure 5.2 based on the result of

Exp. No. 3. The values of K and corresponding r2 values obtained from all experimental

data are summarized in Table 5.3. Higher values of K represent higher removal (R %) of

bitumen. As shown in Table 5.3, the EF of bitumen from oil sand slurries can be

represented by the first order flotation kinetics as the values of r2 is consistently greater

than 0.90. Therefore, Eq. (5.7) can be used to predict the bitumen concentration in oil sand

slurries at time t under the specific operating conditions.

 170

5.2.2 Effect of current density

Effects of the current density on the removal of bitumen versus time are shown in Figure

5.3. The current densities in the experiments varied between 50 and 300 A/m2 while the

other conditions were kept constant. Figure 5.3 shows that the percentage of bitumen

removal (R %) increases with increasing current density up to an optimum value of 150

A/m2. Further increase in the current density causes a decrease in the percentage bitumen

removal. Initially, increasing the current density enhances the generation of hydrogen and

oxygen gases in the flotation cell and therefore, the number of gas bubbles increases;

consequently, the attachment of gas bubbles to bitumen is enhanced, thus more bitumen

drops are brought up by gas bubbles in the froth layer. When the gas volume increases to

a specific threshold at a higher current density (e.g. above 200 A/m2) the bubbles merge

with each other, consequently the number of small bubbles reduces, leading to the decrease

of bitumen removal rate. Figure 5.4 shows that the first order flotation rate constant (K)

increased with increasing current density and decreased after reaching its maximum. The

experimental result reveals that there is an optimum current density for removal of bitumen

from oil sand tailings slurry. This result indicates that increasing the current density after

the optimum level is not feasible for the removal of bitumen. Moreover, higher current

density also increases power consumption. Therefore, it was decided that 150 A/m2 would

be the maximum current density for the removal of bitumen from the MFT slurry.