Chemical Engineering Science 142 (2016) 42–54

Contents lists available at ScienceDirect

Chemical Engineering Science

http://d0009-25

n CorrE-mURL

journal homepage: www.elsevier.com/locate/ces

Mineral beneficiation by ionic microbubble in continuous plantprototype: Efficiency and its analysis by kinetic model

Rajeev Parmar, Subrata Kumar Majumder n

Department of Chemical Engineering, Indian Institute of Technology Guwahati, Guwahati 781039, Assam, India

H I G H L I G H T S

G R A P H I C A L A

� Plant prototype for mineral bene-ficiation by ionic microbubble flota-tion.

� Efficiency of fine particle separationby ionic microbubbles flotation.

� The influence of physicochemicalproperties on fine particle separa-tion.

� Interpretation of separation effi-ciency by development of model.

x.doi.org/10.1016/j.ces.2015.12.00109/& 2015 Elsevier Ltd. All rights reserved.

esponding author. Tel.: þ91 361 2582265 (O)ail address: skmaju@iitg.ernet.in (S.K. Majumd: http://www.iitg.ernet.in/chemeng/SKM.html

B S T R A C T

a r t i c l e i n f o

Article history:Received 4 May 2015Received in revised form25 November 2015Accepted 6 December 2015Available online 11 December 2015

Keywords:CollisionFlotationIonic microbubbleMineral beneficiationRecoveryZeta potential

a b s t r a c t

Microbubbles are miniature gas bubbles of less than 100 μm diameter in liquid. The performance of ionicmicrobubble for fine particle separation is investigated and reported in the paper. The effects of differentoperating variables and physicochemical properties of liquid on the separation characteristics of ionicmicrobubble are enunciated. A phenomenological kinetic model based on collision, attachment anddetachment mechanisms of fine particle is developed to analyze the flotation characteristics of the ionicmicrobubbles. Generalized correlations for flotation rate constant and induction time are also developedbased on the physicochemical properties of microbubble–particle mixture. It is concluded that ionicmicrobubble is highly efficient for removal of fine particles of opposite charge. The findings from thisresearch may be helpful to understand and explain the process better and possibly can be used to modifyand improve the microbubble aided flotation process.

& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Separation of fine particles is widely encountered in mineralindustries. A large sum of operating expenditures of mineralindustries are associated with separation of mineral processes.Therefore the influence of separation process technology onthe profitability is high in mineral industries (Rousseau, 1987).Flotation is most common means of separating the valuablecomponents in this regard. Flotation is beneficial not only tomineral separation, a large variety of chemical species such as

, þ91 361 2584265 (R).er).(S.K. Majumder).

ions, molecules, microorganisms, oil droplets etc. are alsoseparated by this method. They can either be separated fromone another or concentrated from solution (Al-Shamrani et al.,2002; Gaudin et al., 1960; Matis and Mavros, 1991). The coarseparticles are easily separated by the conventional flotationmethod but when the size of mineral particle is in micrometerrange, it is very difficult to separate it. The poor recovery offines mineral by flotation is mainly due to the low probabilityof bubble–particle collision, which decreases with decreasingparticle size (Weber and Paddock, 1983). Conventional flotationare inefficient in encountering collision and attachment of fineparticles and bubbles. For the recovery of fine mineral parti-cles, the flotation cell should have fine bubbles or micro-bubbles suitable to catch these particles (Trahar and Warren,1976). Microbubbles have been extensively used in the mineral

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 43

and liquid effluent treatment, and especially for the flotation ofparticles having size less than 100 mm (Solari and Gochin,1992). Microbubble due to their small size offers a large surfacearea. They experience a low degree of buoyant force, due totheir small size, which gives them a long residence time in theliquid as compared to conventional bubbles (Parmar andMajumder, 2013). The charge on the surface of microbubblehelps it, to become an propitious tool for the fine mineralseparation (Han and Dockko, 1998). According to Jauregi et al.(1997) the potential applications of microbubbles can begrouped into four main zones: (1) flotation for the removal ofbiological and non-biological products; (2) protein recovery;(3) enhancement of oxygen mass transfer; and (4) bior-emediation. Microbubble aided flotation has been widelyemployed in the various fields for the process intensification.They have been used for recovery of proteins (Amiri andValsaraj, 2004; Jarudilokkul et al., 2004; Jauregi et al., 1997;Noble et al., 1998), recovery of microorganism (Hanotu et al.,2012), removal of heavy metal ions from water (Ciriello et al.,1982), removal of dye and pigment (Alves et al., 2006; Royet al., 1992). Cilliers and Bradshaw (1996) used microbubble torecovery of fine pyrites. Shen and Wheelock (2000) reportedthat approximately 85% recovery of fine coal particles can beachieved by using microbubbles. The recovery of fines can beincreased up to 95% using microbubbles with proper chemicalagents (Han and Dockko, 1998). Separation can further beenhanced by using ultrasound with the microbubble (Shibataet al., 2008). Mechanical modification has also been done toimprove this technology for mineral beneficiation (Yi-jun et al.,2009). Microbubble-aided flotation not only increases therecovery it also reduces the frother consumption (Ahmadiet al., 2014). The electrical double layer interactions and theparticle and bubble charge can affect the rate of recovery forfine particles. The electrostatic interactions present betweencharged microbubble and particles of opposite charge may be agoverning force for separation. Fuda and Jauregi (2006) carriedout a detailed study to observe the mechanism of separation ofproteins by ionic microbubbles. They concluded that electro-static interactions were the driving force for the separation.Waters et al. (2008) also used ionic microbubble to separatefine binary mixture of mineral. They found that surface chargeof microbubble can be changed by using surfactants. Theycompared the results obtained with conventional flotationmethod and observed that ionic microbubble has high recoveryover conventional flotation. Recently, Johnson et al. (2009)reported that repulsive long-range interactions are responsiblefor the selective attachment of mineral particles to micro-bubbles in a charge-dependent manner. It is very clear that, theionic microbubble has a potential to intensify recovery inmining industries. When considering the industrial applica-tions of ionic microbubble in mineral separation, it is impor-tant to evaluate the benefits of ionic microbubble based onscientific principles, from an academic standpoint and tocompare microbubble flotation technology with existingtechnology both in terms of its functional quality and effec-tiveness. Thus it becomes very necessary to examine the effectof physicochemical properties of liquid and mineral particlebased on mechanism of ionic microbubble flotation. Thereforethe present work is aimed to study the potential of ionicmicrobubble for the separation of binary fine particle by flo-tation. Furthermore, the effects of surfactants on the recoveryof fine particle are also analyzed. This study may be helpful forthe intensification of the flotation process in the application ofmineral process and technology.

2. Experimental

2.1. Materials

In the present study three mineral particle mixtures such as(i) zinc oxide (ZnO) and silica (SiO2) (ii) copper oxide (CuO) andsilica (iii) aluminum oxide (Al2O3) and silica were used to studythe separation characteristics of ionic microbubble. The surfactantscetyltrimethylammonium bromide (CTAB), sodium dodecyl sulfate(SDS) and Polysorbate 20 (Tween-20) were used to form ionicmicrobubble. The concentrations of CTAB and SDS were variedfrom 5 ppm to 35 ppm. The concentration of Tween-20 was variedfrom 0.015 ml/l to 0.18 ml/l. Preliminary tests carried out on thestability of ionic microbubble showed that 5 ppm of CTAB and SDSeach was sufficient to ensure the dispersion without break downwhen being pumped (Parmar and Majumder, 2015), whereas incase of Tween-20, concentration 0.015 ml/l was sufficient to pro-duce stable microbubble. All the chemicals used in the presentwork had a purity of 498% and were purchased from MERKChemicals (India). The densities of the solution were measuredwith a specific gravity bottle. The surface tension was measured bytensiometer (model K9-MK1, Kruss GmbH, Hamburg, Germany).Previous studies reveals that microbubble suspension is a time-independent non-Newtonian pseudo-plastic fluid and its rheologyis described by Ostwald–De Waele model or Power law model(Parmar and Majumder, 2014; Shen et al., 2008; Tseng et al., 2006)

τ¼ K 0 �dUc

dr

� �n

ð1Þ

where K' and n are constants for a particular fluid. Uc is micro-bubble–particle mixture velocity and r is the radial distance. Theconstant K' is known as the consistency of the fluid and n is knownas flow behavior index. The parameter n and K' are independent ofgeometry of the column (Larmignat et al., 2008). The values of nand K' are calculated experimentally in a horizontal pipe from thewall shear stress (τ) and the apparent shear rate (γa) as (Larmignatet al., 2008)

τ¼ αγna ð2ÞThe parameter α and the flow consistency index (K') of the

microbubble–particle mixture are related as (Larmignat et al.,2008)

α¼ K 0 3nþ14n

� �n

ð3Þ

The wall shear stress and apparent shear rate can be calculatedfrom the volumetric flow rate of mixture (Qm) and pressure drop(ΔP) in the pipe according to the following relations (Larmignatet al., 2008)

τ¼DpΔP4Lp

ð4Þ

γa ¼32Qm

πD3p

ð5Þ

where Dp is diameter of pipe and Lp is length of pipe. The effectiveviscosity of microbubble–particle mixture (me) in the flotationcolumn is defined as the ratio of the shear stress at the wall to theaverage shear rate which was calculated by

μe ¼ K 0 8Uc

Dc

� �n�1

ð6Þ

All the experiments in the present work were carried out at2571 °C. The physical properties of liquid phase are listed inTable 1.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5444

2.2. Zeta potential measurement

The zeta-potential of particles in surfactant solution (surfactantand water) was measured using a zeta potentiometer (Model-Delsa Nano C, Beckman Coulter, Nyon, Switzerland) from theelectrophoretic mobility of the particle. Each data point for zeta-potential was taken as an average of three measurements at roomtemperature. In the present study all the experiments were

Table 1Physical properties of system measured at 2571 °C.

Type of fluid–par-ticle mixture

Concentration inwatera

Density (kg/m3) Surface tension(mN/m)

SDS-1 5 999.73 69SDS-2 10 989.76 67.00SDS-3 15 985.79 65.20SDS-4 20 981.02 63.00SDS-5 30 978.84 62.00CTAB-1 5 994.74 69.20CTAB-2 10 989.77 66.37CTAB-3 20 987.02 63.77CTAB-4 50 984.84 60.75CTAB-5 100 981.21 59.83Tween-20-1 0.25 1000.08 64.03Tween-20-2 0.50 999.10 57.66Tween-20-3 1.0 991.028 52. 36Tween-20-4 2 988.94 48.13Air – 1.18 –

a The units are in ppm for SDS and CTAB; in mg/l for Tween-20.

Fig. 1. Size distributions of di

conducted within the pH range 8–9. So variation of zeta potentialof particles and bubble with the pH was neglected.

2.3. Microbubble generation

Pressurized dissolution method was used to produce themicrobubbles in the surfactant solutions. In this method, mixtureof gas and liquid were mixed and transported to pressure cham-ber, where the gas was dissolved at the saturation concentration. Ahigh pressure was built up in the chamber by adjusting the outletflow rate. The water with oversaturated air was released to lowatmospheric pressure. Microbubbles were produced by flashingthis saturated liquid through microbubble output control valveinto the expelled liquid (Parmar and Majumder, 2013). Themicrobubble dispersions were prepared in a 7.5�10�2 m3 vessel.The dispersion of microbubbles in water were continuously recy-cled through the vessel. The void fraction of microbubbles gener-ated by the pressurized dissolution method was determined bythe concentration of dissolved gas that depends on the pressure atthe dissolution section. The pressure at the dissolution section canbe varied manually to produce microbubble within the range of 1–100 μm.

2.4. Particle size and microbubble size measurement

Laser particle size analyzer (Malvern Instruments Ltd., model-APA 2000, Malvern UK) was used to calculate the microbubble andparticle size. The laser diffraction method delivers rapid bubblesize distributions over milli to nanometer ranges. The particle size

fferent mineral particles.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 45

distributions measured by particle size analyzer is shown in Fig. 1.The microbubble size, produced in a liquid depends on the phy-siochemical properties of liquid. A significant change in the surfacetension of liquid can change the bubble size distribution profileand influence the bubble stability. In the present work, it wasfound that the microbubble size generated in the particle–liquidmixture by the microbubble generator is mainly affected by sur-face tension of the fluid. The size of microbubble produced can beexpressed as

dmb ¼ 0:43σ3:4 ð7Þ

where dmb is the microbubble diameter and σ is the surface ten-sion of liquid. A typical size distribution of microbubble at differ-ent Tween-20 concentration in the liquid is shown in Fig. 2.

001010

2

4

6

8

10

12

14

16

18

db (μm)

Volu

me

(%)

Ct=0.015 (mg/l)

Ct=0.031 (mg/l) Ct=0.062 (mg/l) Ct=0.125 (mg/l)

Fig. 2. Typical microbubble size distribution in Tween-20 surfactant.

Fig. 3. Schematic of experimental setup. Legend: -C: air compressor; FC: flotation columanalyzer; MBG: microbubble generator; Mo: microbubble generator output; Mi: microburotameters; S: froth sample collection; Tm: thermometer; T: titration unit; V1, V2: contr

2.5. Experimental setup and procedure

The schematic of the plant prototype in which the experimentswere carried out is shown in Fig. 3. The experimental systemincludes a flotation column, a microbubble generator, a gas com-pressor and other accessories such as rotameters, control valves,thermometer, etc. The separation system is composed of themicrobubble flotation column. The flotation column was made ofplexiglass having an internal diameter of 0.2 m and a height of0.6 m. The top portion of the column is made hooded for collectionof froth. Both inlet and outlet of the column were at the lower endthrough which microbubble generator was connected. The airused to create microbubbles was supplied to the generator from acompressor controlled by a valve. The microbubble particle mix-ture flow rate was measured by using rotameters for which theflow was maintained by using control valves. Equal mass of finemineral particle and silicon dioxide (10 g each) were used a asflotation feed. The experiments were conducted at six differentmixture circulation velocities (Uc) in the range of 0.24–0.36 m/s.The circulation velocity in the flotation column is determined byQm/(Ac/2) where Qm is the microbubble particle mixture volu-metric flowrate measured by rotameter and Ac is the cross-sectional area of flotation column. The mixture circulation velo-city is taken on the basis of half of cross-sectional area of column,assuming that the microbubble particle mixture flows up throughhalf of column cross-section and recirculates. The froth and floatedparticles were removed from the top of the column at every threeminutes interval, after the initiation of experiment. The collectedparticles were dried in an oven (Reico Equipment and InstrumentPvt. Ltd., model-ROV/DC, Kolkata India), till the equilibriummoister content is reached and then weighed. The collected sam-ples contained both silica and the mineral particles. The amount ofmineral particle from the mixture was determined by thefollowing way:

n; CB: control board; Gi: Microbubble generator gas inlet; LPSA: Laser particle sizebble generator input; M1, M2: manometers; O: electric oven; PC: computer; R1, R2:ol valves; W: weighing balance.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5446

2.5.1. Determination of ZnOInitially a known volume of 2 N H2SO4 was poured in a beaker

containing the dried mixture of SiO2 and ZnO. Silicon dioxide isinert to H2SO4 whereas ZnO readily reacts with H2SO4 to form zincsulfate and precipitates. The unreacted H2SO4 acid was thenseparated from the mixture.

ZnOþH2SO4-ZnSO4↓þH2O ð8ÞThe volume of unreacted H2SO4 was determined by acid base

titration. The amount of ZnO in mixture was then determined fromthe stoichiometry analysis.

2.5.2. Determination of Al2O3 and CuO.The dried mixture of CuO/Al2O3 and SiO2 was first reacted with

2 N HCL. Silicon dioxide is inert to HCl whereas CuO/Al2O3 readilyreacts with HCl to form respective chlorides and precipitates.

CuOþ2HCl-CuCl2↓þH2O ð9Þ

Al2O3þ6HCl-2AlCl3↓þ3H2O ð10ÞThe volume of unreacted HCl was determined by acid base

titration. The amount of CuO/Al2O3 in mixture was determined bythe stoichiometry analysis. Each experiment was replicated threetimes and the results showed a maximum variation of 2% of themean value. Finally, the percentage removal efficiency or recovery(Rc) was calculated as per Eq. (11)

Rc ¼Mr

Mf� 100 ð11Þ

where Mr represents mass of particle recovered and Mf denotesmass of feed.

3. Results and discussion

3.1. Effect of mixture velocity on recovery of mineral particles

Flotation of fine particles by using microbubble is very complexphysicochemical process. The number of variables which can affectmineral recovery may be classified broadly into the physical andthe chemical. In the present work, effect of these variables areexamined in detail. Fig. 4 shows the effect of mixture circulationvelocity on the flotation recovery, at fixed surfactant concentra-tion, for all the three mineral particles. It is observed that for allthe particles, the flotation recovery increases with the increase inmixture velocity. The particle microbubble collision depends on

Fig. 4. Effect of circulation velocity of mixture on the recovery of particles.

the hydrodynamics of flow, which is mainly governed by mixturevelocity. The turbulence in the column increases due to increase inmixture velocity. The mixing characteristics of microbubble–par-ticle mixture in the column also increases with increase in tur-bulence in the column. Mixing increases the probability of bubble–particle collision. As the particle come nearer to the microbubbles,it colloids to the bubble. After colliding with a rising microbubble,the particle attaches on the bubble surface and forms a stableparticle–bubble aggregate. However, all the particles collidingwith air bubbles do not result in flotation. Only the hydrophobicparticles adhere to the surface of bubbles. The attached particle onthe microbubble moves upward due to buoyant force acting on themicrobubble and collected from the top of the flotation column.Therefore the flotation recovery increases with increase in mixturecirculation velocity. Particle detachment may also occurs whendetachment forces exceed the maximum adhesive forces. Oneprobable source of adhesive forces is bubble oscillations caused byparticle and bubble collisions, which is quite often for coarseparticles. It is also observed from Fig. 4 that for same surfactantand for its same concentration, the flotation recovery is differentfor different particle. This is mainly due to difference in physicalproperties such as diameter, density, character of the surface, etc.of particles. These properties affect the collision process betweenthe particle and microbubble. Therefore the particle shows dif-ferent recovery for same surfactant concentration.

3.2. Effects of surfactants on flotation recovery of ZnO

The concentration of surfactant is another important aspect offlotation that affects flotation performance. The effect of surfactantconcentration on the recovery of ZnO is shown in Fig. 5. It is seenthat increase in surfactant concentration increases the ZnOrecovery. Surfactant not only influence the density of microbubble,they also provide stability to microbubble (Feng et al., 2009). Theprobability of collision between a particle and a microbubble isdefined as the fraction of particles that collide with a risingmicrobubble in its path, significantly affected by surfactant. Theprobability of collision increases with increase in the particle sizeor decrease in the microbubble size. Addition of surfactant sig-nificantly decreases the microbubble size. For a microbubblegenerated in the liquid, the probability of collision with particles isvery high because of its small size, which is in part responsible forhigher recovery at a given frother concentration. Addition of sur-factant provides three main functions: (i) addition of surfactant inwater alters the interfacial surface tension. It was observed from

Fig. 5. Effects of surfactants on the recovery of ZnO.

Fig. 6. Effect of concentration of surfactants on the zeta potential of ZnO.

Fig. 7. Effect of surfactants on the recovery of CuO particles.

0 10 20 30 40

-15

-10

-5

0

5

10

15

Cc (ppm), Cs (ppm), Ct × 102 (ml/l)

ζ (m

V)

CTAB SDS Tween-20

Fig. 8. Effects of concentration of surfactants on the zeta potential of CuO particles.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 47

the previous studies that the microbubble size decreases withincrease in surfactant concentration. Surfactant lowers the surfaceenergy of the gas liquid interface and thus helps to create smallerbubble, even low concentrations of surfactants in the liquid arecapable of reducing bubble sizes. The population of microbubblealso increases with increasing surfactant concentration (Coutoet al., 2009; Parmar and Majumder, 2015; Xu et al., 2009). There-fore, the microbubble population in flotation column significantlyincreases with increasing surfactant concentration, which assist toattach more particles and enhances the recovery, (ii) the surfactantreduces the microbubble size, thereby increase collision prob-ability. The collision probability of bubbles increases with decreasein bubble diameter (Reay and Ratcliff, 1973). Ahmed and Jameson(1985) also found that the flotation rate was very strongly affectedby the bubble size. They reported that reduction of bubble sizefrom 655 mm to 75 mm results an increase in rate constant up toone hundred-fold, (iii) the third important function of surfactant isthat they control the surface characteristics of bubble. Interfacialrheological properties, stability and the strength of charge on thesurface of microbubble depends on the surfactant (Saulnier et al.,1996; Shah et al., 1978; Xu et al., 2009). The surfactant SDS is ananionic in nature and CTAB is a cationic surfactant. The surfactantTween-20 is non-anionic in nature. Therefore the microbubbleformed by SDS and CTAB carries anionic and cationic chargerespectively. It is also observed from Fig. 5 that the recovery of ZnOis different for different surfactants. In case of Tween-20, at fixedmixture circulation rate, the recovery of ZnO just increase from60% to 63% as surfactant concentration increases from 0.016 ml/l to0.18 ml/l. However a higher recovery was obtained in SDS ascompared to Tween-20. As the SDS concentration increases from5 ppm to 30 ppm the recovery increases from 63% to 68%. Themaximum recovery of concentrate (ZnOþSiO2) as well as ZnO isobtained in case of CTAB. At a fixed circulation rate, as the sur-factant concentration increases from 5 ppm to 30 ppm, therecovery of ZnO increases from 69% to 84.1%. Naturally micro-bubbles carry negative charge on its surface (Takahashi, 2005).However this charge be altered by adding a suitable surfactant(Yoon and Yordan, 1986). As the ionic microbubbles pass throughthe feed, oppositely charged mineral particles are attracted easilydue to electrostatic forces. The zeta potential of ZnO particle isapproximately �10 mV. Addition of surfactant significantly influ-ences the zeta potential of particle as shown in Fig. 6. Addition ofSDS and Tween-20 in the feed increases the negative charge onthe ZnO surface, which is not worthwhile, as it does not create anysignificant attraction between the ionic microbubbles and the ZnOparticles. The development of charge at the surface of particle,affects the distribution of ions in the interfacial region. Con-sequentially an increase in concentration of opposite ions results

in close to the surface. Thus an electrical double layer existsaround each particle. The microbubbles formed with CTAB carrynegative charge which easily attract the opposite charged ZnOparticles. Therefore the recovery of ZnO is highest with CTAB ascompared to SDS and Tween-20.

3.3. Effects of surfactants on flotation recovery of CuO

The nature and magnitude of charge on the surface of particleas well as on the microbubble are significant parameters for ionicmicrobubble flotation. The effects of addition of different surfac-tants on the CuO recovery is shown in Fig. 7. It is observed thatrecovery of CuO in the concentrate increases with increase insurfactant concentration, which was expected. However, themaximum recovery of copper oxide was achieved by using SDS,compared to Tween-20 and CTAB. CTAB as a cationic surfactantprovides positive charge on microbubble whereas SDS and Tween-20 provide negative charges on the surface of microbubble. Thezeta potential of CuO particle is positive. It is changed by additionof surfactant as shown in Fig. 8. Negative charged microbubbleleads to high recovery of CuO particles. This high recovery indi-cates a higher attachment of CuO particles, and emphasizes thepotential of separation using charged microbubbles.

The maximum recovery of CuO with SDS is approximately 80%and the surfactant consumption of SDS up to maximum recovery isapproximately 35 ppm, which is much less than the critical micelle

Fig. 9. Effect of concentration of surfactants on the zeta potential of SiO2.

Fig. 10. Effects of surfactants on the recovery of Al2O3 particles.

Fig. 11. Effect of concentration of surfactants on the zeta potential of Al2O3.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5448

concentration. So ionic microbubbles significantly reduces sur-factant consumption. This can be associated to the increase in thecontact angle and the contact area between a microbubble and aparticle in the presence of microbubbles (Chipfunhu et al., 2011).Collector increases the particle hydrophobicity, thereby accel-erating the liquid film drainage and particle–bubble attachment(Ralston et al., 2002). The increase in the contact angle alsoincreases the attachment force between the microbubble andparticle, leading to decreased probability of particle detachmentfrom the bubble (Fan et al., 2010). In case of ZnO and SiO2 mixturethe concentration of concentrate (both SiO2 and ZnO) increaseswith increase in CTAB concentration. However in case of CuO andSiO2 mixture with SDS, the grade recovery of CuO was higher thansilica. The zeta potential of silica is found to be negative as shownin Fig. 9. The zeta potential of silica particle in the feed mixture isapproximately �41 mV. It is also seen that the zeta potential ofsilica did not change significantly on addition of SDS as well asTween-20, which clearly indicates that negative charge does notadsorb on the negative silica surface. However adsorption of CTABsignificantly alters the zeta potential of SiO2. The zeta potentialincreases in magnitude, until there was a charge reversal after theaddition of approximately 20 ppm of CTAB in feed. This indicatesthat the CTAB was adsorbed onto the negatively charged surface ofthe silica, which results in high silica recovery.

3.4. Effects of surfactants on Al2O3 recovery

The flotation recovery of Al2O3 as function of surfactant con-centration is shown in Fig. 10. It is observed that the SDS andTween-20 are found to be less effective as compared to CTAB forAl2O3 recovery. However, the flotation recovery in all the threesurfactants increases with increasing concentration. The max-imum recovery of Al2O3 achieved in the present work is 73.7%with CTAB, which is less than ZnO (84.1%).

There are a large number of physical variables which can affectthe recovery of particles by microbubbles. The maximum recoveryof particle depends upon its floatability and other physiochemicalproperties. (Trahar and Warren, 1976). Low negative zeta potentialof Al2O3 also affects its recovery. The zeta potential of Al2O3 par-ticle is negative and has low magnitude as shown in Fig. 11. Thezeta potential of Al2O3 in water is approximately �3.6 mV whichis low in magnitude as compared to CuO and ZnO. The attachmentprobability of particle depends on particle size, surface charge andsurface roughness (Krasowska and Malysa, 2007). Although theduration of electrostatic attraction between bubbles and particlesis very short, a large difference between zeta potentials of bubbleand particle can considerably increase the recovery efficiency. Thedifference in potential of bubble and particle is least for Al2O3

particle. Therefore the recovery of Al2O3 is least. Collins andJameson (1977) reported that electrostatic interactions throughoverlapping of diffuse double layers between microbubbles andparticles affect the rate of flotation and therefore particle collec-tion. Ralston (1983) also interpreted that surface forces has influ-ence on particle trajectory. It is the consequence of thick wettingfilms stabilized on hydrophobic surfaces by long-range electro-static forces (Ahmadi et al., 2014). These results indicate thathydrophobic as well as electrostatic forces are responsible forparticle collection by ionic microbubbles.

3.5. Variation of recovery of particle with time

It is observed that maximum fractional recovery profiles for allthe three particle are quite similar to each other. A typical varia-tion of recovery of particle with time at fixed mixture circulationvelocity and surfactants concentration is shown in Fig. 12.

It is observed that approximately 60% of the particles arerecovered within 11 min of flotation and in the later stage thefractional recovery is low. The fractional recovery decreases withtime. In the initial phase the mineral particle shows a higherattraction towards the ionic microbubble than the later phase.Since the density of microbubble at high surfactant concentrationis very high so it becomes very easy for the microbubble to attachthe particle. In the initial phase the surface of particle is

Fig. 12. Variation of recovery of particles with time.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 49

uncontaminated with adsorbed surfactant but as the time passesthe opposite charged surfactant molecules adsorb on the surface ofparticle that causes a reduction in attraction and attachment.

4. Model for microbubble flotation

Flotation column operates on three main distinct processesnamely: (i) aeration, where air bubbles are introduced into theflotation column; (ii) mixing, where bubbles and particles arewell mixed to increase bubble–particle interaction; and finally(iii) separation, where bubbles and bubble–particle aggregates areallowed to separate from the bulk mixture and skimmed away.Many fundamental models have been developed to mark out theflotation kinetics, predominantly related to the collection zone interms of hydrodynamics. A comprehensive description on funda-mentals of the flotation process modeling for column flotation isreported by Finch and Dobby (1990), Tuteja et al. (1994) and Daiet al. (2000). Kinetic model is widely adopted for development ofcontrol strategies in flotation column. Kinetic models involve thechemical reactor analogy and consider flotation process as areaction between bubbles and particles (Jameson et al., 1977). Inthe present work the flotation of fine mineral particles by micro-bubbles is modeled as a first-order rate process. The kineticequation describing the rate of removal of the number of particles,for the microbubble–particle encounters can be expressed as(Bloom and Heindel, 1997)

dNp

dt¼ �KNp ¼ ZEcEaEs ð12Þ

where Np is the number densities of floatable mineral particlesavailable for attachment. K denotes the rate constant for particle–microbubble flotation. The parameter Z is related to the particle–microbubble collision frequency dependent on the size of theparticles and microbubbles, and hydrodynamics of the flotationpulp. Ec, Ea and Es are particle–microbubble collision, adhesion andstability efficiencies respectively. The parameter Z can be expres-sed as (Pyke et al., 2003)

Z ¼ 5NmbNpdmbþdp

2

� �2 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiV

2mbþV

2p

qð13Þ

where V2mb

� �1=2and V

2p

� �1=2are the root mean square velocity

(rms) of the particles and microbubbles relative to the fluid velo-city, respectively. Nmb denotes microbubble density. The number

density of microbubbles in the column can be calculated as

Nmb ¼6εgπd3mb

ð14Þ

where εg denotes the gas holdup. In the present work as the

dmb»dp, V2p

� �1=2can be neglected. So Eq. (13) becomes

Z ¼ 30Npεgπd3mb

dmbþdp2

� �2

V2mb

� �1=2ð15Þ

The rms relative velocities of the microbubbles can be calcu-lated as (Schubert, 1999)

V2mb

� �1=2¼ 0:33ε4=9d7=9mb

ν1=3Δρρf

!2=3

ð16Þ

where Δρ is the density difference between the particle (ρp) andfluid (ρf), ν is kinematic viscosity of fluid and ε is the energydissipation rate per unit mass. It can be calculated as (Wu andPatterson, 1989)

ε¼ 0:85U

2cxþU

2cyþU

2cy

2

0@

1A

3=2

L�1 ð17Þ

where Uc is the microbubble–particle mixture circulation velocity.L is a measure of the flow length scale. The length scale can betaken as 0.07 Dc (Pope, 2000). Therefore Eq. (17) can be expressedas

ε¼ 12:143U

2cxþU

2cyþU

2cz

2

0@

1A

3=2

D�1c ð18Þ

Thus the expression of Z of Eq. (15) can be written by incor-porating Eqs. (16) and (18) as

Z ¼ 30Npεgπd3mb

dmbþdp2

� �2d7=9mb

ν1=3Δρρf

!2=3U

2cxþU

2cyþU

2cz

2

0@

1A

3=2

D�1c

0B@

1CA

4=9

ð19ÞSubstituting the expression of Z from Eq. (19) in Eq. (12), one

gets the expression of K as

K ¼ 30εg

πd3mb

dmbþdp2

� �2d7=9mb

ν1=3Δρρf

!2=3U

2cxþU

2cyþU

2cz

2

0@

1A

3=2

D�1c

0B@

1CA

4=9

EcEaEs

ð20ÞThe root mean square velocity of microbubble–particle mixture

is assumed to equal to root square mean of circulation velocity,Hence Eq. (20) becomes

K ¼ 30εg

πd3mb

dmbþdp2

� �2d7=9mb

ν1=3Δρρf

!2=3U

2c

2

!3=2

D�1c

0@

1A

4=9

EcEaEs

ð21ÞThe percentage recovery of particle (Rc) at any time t can be

expressed as

Rc;t ¼ Rmax 1�exp �Ktð Þð Þ ð22Þ

4.1. Collision efficiency

The collision efficiency is the ratio of the number of particlescolliding with the bubble per unit time to the number of particlesswept across the projected area of the bubble per unit time(Weber and Paddock, 1983). Dai et al. (2000) presented a critical

Table 2Contact angles of metal oxide in air–water at standard condition taken for calcu-lation of the model parameter.

Metal oxides Contact angle (deg) Reference

Al2O3 61.2 Poljacek et al. (2008)SiO2 79 Rulison (2000)CuO 90 McDonald and Cui (2011)ZnO 104.5 Subedi et al. (2011)

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5450

review of the various models existing in the literature for thecalculation of the collision efficiency between particles and singlerising gas bubbles. In most particle–bubble collision models, majorconsideration has been given to bubbles with completely immo-bilized surfaces since it is believed that during flotation, the bub-ble surface is completely retarded by the presence of surface activeimpurities from the water. For mobile bubble surface, generalizedSutherland equation (GSE) and Weber and Paddock model canwell predict the collision efficiency. Weber and Paddock (1983)proposed a model for the collision of spherical particles. Theyassumed that the particles were very small and the hydrodynamicinteraction between the particles and the fluid was negligible. Thestreamlines of the fluid could be characterized by the Stokesstream function and the stream functions could be approximatedby a Taylor series. According to them the collision efficiency can bedissected into two components: one relating to the particle set-tling velocity due to gravity and the other relating to the fluidvelocity at the bubble surface. The former component is termedthe gravitational effect and the latter is the interceptional effect.Both effects depend on the bubble Reynolds number (Ren) andcalculated by solving the Navier–Stokes equations. The collisionefficiency according to Weber and Paddock model can be expres-sed as

Ec ¼ EcgþEci ð23Þ

The collision efficiency due to the gravitational effect (Ecg) isexpressed as (Weber and Paddock, 1983)

Ecg ¼Up

1þUp1þ dp

dmb

� �2 !

sin 2 θc ð24Þ

where Up is particle velocity, θc is denoted by the maximum col-lision angle above which no collision is possible. The value ofmaximum collision angle depends on the bubble Reynolds numberand can be determined by the following equations, obtained byempirical curve fitting to collision data (Woo, 1971)

θc ¼ 78:1�7:37 log Renð Þ For 20oReno400 ð25Þ

θc ¼ 98�12:49 log 10Renð Þ For 1oReno20 ð26Þ

θc ¼ 90�2:5 log 100Renð Þ For 0:1oReno1 ð27Þ

where Ren denotes the non-Newtonian Reynolds number. Thenon-Newtonian Reynolds number can be expressed as

Ren ¼dnmbU

2�nc

8n�1K 04n

3nþ1

� �n

ð28Þ

The collision efficiency due to the interceptional effect (Eci) canbe expressed as (Weber and Paddock, 1983)

Eci ¼ 1:5 1þ 316

� �Ren

1þ0:249Ren

� �� �dpdmb

� �2

For Reno200 ð29Þ

4.2. Stability efficiency

The stability efficiency addresses the stabilization/destabiliza-tion of a microbubble–particle aggregate. According to Schulze(2003), the acceleration (force) that determines the detachment ofa particle from the bubble is dependent on the intensity of tur-bulence in the flow. The capillary and hydrostatic forces constitutethe attachment forces while the forces of gravitation, buoyancyand capillary pressure in the gas bubble constitute the detachmentforces. The sum of these forces is zero at equilibrium (Duan et al.,2003). The expression for stability efficiency can be written as

(Schulze, 2003)

Es ¼ 1�exp 1� 6σ sin ω sin ωþθ �� ��

d2p Δρgþρpac� �

þ1:5dp sin ωð Þ2f dmbð Þ

0@

1A ð30Þ

where σ is the surface tension,ω refers to the location of a particleat the liquid–vapor interface (ω¼180°�θ/2), g is the gravitationalconstant and ac is the particle centrifugal acceleration in a tur-bulent flow field which depends on the level of turbulence in theflotation column. In the present work, the variation of contactangle with surfactant concentration is ignored. The contact angleof metal oxides in air–water at standard condition taken for cal-culation of the model parameter is given in Table 2. The function f(dmb) can be expresses as (Schulze, 2003)

f dmbð Þ ¼ 4σdmb

�dmbρf g� �

ð31Þ

The particle centrifugal acceleration (ac) for the particle smallerthan the microbubble can be approximated as

ac ¼ 1:9ε2=3

d1=3mb

ð32Þ

4.3. Attachment efficiency

The attachment of a hydrophobic particle to a bubble is one ofthe significant phase of particle–bubble interaction in mineralflotation. Particle–bubble attachment occurs when the particle–bubble contact time is longer than the induction time (Ti). Theinduction time is defined as the time for the liquid film betweenthe particle and the bubble to thin and rupture and for the three-phase line of contact to expand until an equilibrium value isobtained. The contact time is linked to the particle–bubble colli-sion. If a particle impacts on the bubble surface with an adequatekinetic energy so as to cause substantial distortion of the bubblesurface, the colliding particle rebounds from the deformed surfacedue to the elastic energy of the deformed part of the surface. Thenumber of attachment models are limited compared to the colli-sion models. The attachment efficiency (Ea) is defined as thefraction of all colliding particles that reside on the bubble for atime greater than the induction time (Dobby and Finch, 1987).After the collision, the particle slides over the bubble and attach-ment occurs when the intervening liquid film thins and ruptures(Dobby and Finch, 1987). Consequently, particles with a slidingtime greater than induction time were considered to haveattached on bubbles. The model of Dobby and Finch (1987)requires the knowledge of the distribution of particle on thebubble surface, particle collision angles, the maximum angle ofcontact between the particle and bubble and the particle slidingvelocity. Later on the model was modified to improve theapproach. Based on the modification, the attachment efficiency forparticle–microbubble can expressed as (Dai et al., 1999)

Ea ¼ sin 2 θa

sin 2 θt

ð33Þ

Fig. 13. A typical outline to estimate the induction time from the experimentalvalues of recoveries.

Fig. 14. Variation of induction time of ZnO particle with mixture circulationvelocity.

Fig. 15. Variations of the induction with Reynolds number at different surfactantconcentrations.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 51

where θt is the maximum possible collision angle of the particle onthe surface of the bubble which is given by (Dukhin, 1982)

θt ¼ arcsin 2β 1þβ2� �1=2

�β� �� 1=2

ð34Þ

where β is dimensionless number, which is a measure of therelative importance of the interceptional and inertial contributionsto the collision process in the generalized Sutherland equation andis defined as (Dai et al., 1998)

β¼ 4Esu9Ka

ð35Þ

where Esu is the collision efficiency calculated by (Sutherland,1948), (Esu¼3dp/dmb). The parameter Ka is defined as

Ka ¼2Uc ρp�ρf

� �d2p

9νdmbð36Þ

The parameter θa is the adhesion angle at which the particlecollides with the bubble. Its sliding time (Ts) is equal to theinduction time (Dai et al., 1998). So the angle relates the slidingtime and induction time to the attachment efficiency. The adhe-sion angle under potential flow conditions can be expressed as(Dobby and Finch, 1987)

θa ¼ 2arctanexp Ti

2 UspþUp þUb

dpdp þdmb

� �3dpþdmb

264

375 ð37Þ

In the present study, the induction time (Ti) is calculated fromthe best fit of Eq. (22) and experimental values of recoveries.Fig. 13 shows a typical outline to estimate the induction time fromthe experimental values of recoveries for ZnO and SiO2 mixturewith CTAB.

4.4. Interpretation on Induction time

Sven-Nilsson (1935) introduced the concept of induction timewho measured induction time by moving a bubble towards andthen away from a flat mineral surface. The author consideredinduction time to be the minimum contact time for successfulthinning of the intervening liquid film to critical thickness wherefilm rupture occurs. It has been reported that induction time is afunction of many physical parameters. Yoon and Yordan (1991)observed a power law dependence between measured inductiontime and particle size, which was also coherent with theoreticalanalysis. The induction time depends not only on particle size but

also on many other variables such as, the bubble size, the surfacetension, disjoining pressure, and the viscosity of the continuousphase (Li et al., 1990). The present results shows that inductiontime depends on the concentration and type of surfactant, mineralparticle as well as the mixture velocity. A typical variation ofmicrobubble–particle induction time with mixture circulationvelocity at different surfactant concentrations for ZnO particles inCTAB is shown in Fig. 14. It is seen that induction time reduces asthe mixture velocity increases. Increase in mixture velocity redu-ces the time for bubble–particle adhesion, leading to reduction inthe induction time. Ye et al. (1989) also reported the dependencyof induction on the velocities of the bubble approaching the par-ticle. Another important conclusion that can be drawn from Fig. 14is that, induction time decreases with increasing surfactant con-centration, while the flotation recovery was found to be increasedconcomitantly. This may be attributed to increase in adsorptiondensity at the mineral surface with increasing surfactant con-centration. Both contact angle and induction time are stronglycontrolled by surfactant concentration. However, the contact angleis generally an equilibrium measure of hydrophobicity while theinduction time is a kinetic measurement of the hydrophobicity.Particles having a shorter induction time will be more easilycaptured, and they are said to have a higher selectivity. Based onthe present experimental data, the induction time has been

Fig. 16. Variation of floatation rate constant of CuO particle with mixture circula-tion velocity.

Fig. 17. Variations of the flotation rate constant with Reynolds number at differentsurfactant concentrations.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5452

correlated in terms of the operating variables by the dimensionalanalysis. The correlation is made by fitting present experimentaldata with the help of multiple regression analysis by MicrosoftExcel 2013, which can be expressed as

Ti ¼ 1:98� 103 dmb

Uc

� �1:172 dpdmb

� ��0:592

Ca0:405Re�0:434n ð38Þ

where Ca denotes Capillary number (mfUc/σ). The correlationcoefficient and standard error of Eq. (38) are found to be 0.993 and0.051 respectively. A typical parity plot for the comparison ofexperimental and predicted values of induction time of mixturefor Al2O3 particles is shown in Fig. 15. It is clear that proposedcorrelation can predict induction time well within the followingranges of operating variables: 4.51�10�5rdmb/Ucr2.33�10�4,8.03�10�2rdp/dmbr92.20�10�2, 3.81�10�3rCar11.22�10�3 and 2.47rRenr18.24.

4.5. Interpretation on flotation rate constant

The scale-up of the column requires the knowledge of flotationconstant. A large number of variables may affect the ultimateperformance or rate of flotation in column. Fig. 16 shows the effectof mixture circulation velocity on flotation rate constant for thebeneficiation of CuO by SDS. It is seen that flotation rate constantstrongly depends on the circulation velocity and surfactant con-centration. Increase in the circulation velocity significantlyincreases the collision. It is also observed that surfactant con-centration has a significant effect on the rate constant. Surfactantincreases the rate constant in two ways. First, it reduces themicrobubble size which helps to increases the particle bubblecollision (Parmar and Majumder, 2015). It also increases the ionicstrength of microbubbles (Yoon and Yordan, 1986). Ionic strengthhas a significant role to play in determining the adsorption ofsurfactant on the mineral as well as on the microbubble due toincrease in electrical double-layer. Based on the present experi-mental data, the flotation rate constant has been correlated interms of the affecting variables by the dimensional analysis, whichcan be expressed as

K ¼ 5:3� 10�10 dmb

Uc

� �dpdmb

� ��0:064

Ca�1:04Re0:289n ð39Þ

The correlation coefficient and standard error of Eq. (39) are0.989 and 0.052 respectively. A typical parity plot of the correla-tion proposed for the rate constant with the experimental valuefor the beneficiation of Al2O3 by CTAB is shown in Fig. 17. It was

found that the developed correlation fitted well within the rangesof variables: 4.51�10-5rdmb/Ucr2.33�10�4, 8.03�10�2rdp/dmbr92.20�10�2, 3.81�10�3rCar11.22�10�3 and 2.47rRenr18.24.

5. Conclusion

The technical feasibility of an innovative method for finerecovery was investigated. The present work shows that fineparticles can be significantly recovered by using microbubbles. Theresults showed that the charge on the surface of microbubble ishighly promising in separating opposite charged particles. Therecovery of mineral particle was found to be dependent on sur-factant concentration, size of microbubble and particles, zetapotential of microbubble, nature of surface potential of bubble andmicrobubble–particle mixture circulation velocity. It is alsoobserved that the separation efficiency of microbubble increaseswith increase in mixture circulation velocity. The addition of sur-factant in the mixture intensify the removal efficiency. Therecovery of ZnO and Al2O3 particles was maximum with CTAB,whereas the Tween-20 and SDS were found be less effective forZnO and Al2O3, due to similar surface charge. As the CTAB con-centration increases from 5 ppm to 30 ppm, the recovery of ZnOincreases from 69% to 84.1%, whereas for the Al2O3 particle therecovery increases from 66.1% to 73.7%. In case of CuO particles,the SDS and Tween-20 were found to be more effective than CTAB.The maximum recovery of CuO with SDS and Tween-20 are 80%and 60% respectively. The SDS microbubbles possessed highermagnitude of zeta potential than the Tween-20 microbubble,which causes the higher recovery. The zeta potential of SiO2 par-ticles was found to be independent of SDS and Tween-20 con-centration within the experimental range. However in case ofCTAB, a reversal of charge with increase in concentration of CTABwas observed. A flotation model that includes the contributionsfrom the efficiencies of collision, attachment and stability betweenparticles and microbubbles was used to calculate the flotation rateconstants. It is observed that rate constant is significantly influ-enced by the physicochemical properties of the liquid and parti-cles. The presence of a hydrophobic surface force necessarilyresults in particle collection by a microbubble. Other surface forcesalso contribute in fine particle separation. The application ofmicrobubble in fine particles separation may be useful forseparation of other type of fine particles efficiently based on theinsight of the present study.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–54 53

Notation

Ac Column cross-sectional area (m2)ac Particle centrifugal acceleration (m2/s)Cc Concentration of CTAB (ppm)Cs Concentration of SDS (ppm)Ct Concentration of Tween-20 (ml/l)Dc Column diameter (m)dmb Microbubble diameter (m)dp Particle diameter (m)Ec Particle–microbubble collision efficiency (–)Eci Collision efficiency due to the interceptional effect (–)Ecg Collision efficiency due to the gravitational effect (–)Ea Particle–microbubble attachment efficiency (–)Es Particle–microbubble stability efficiency (–)g Acceleration due to gravity (m2/s)K0 Consistency of fluid (Pa sn)K Flotation rate constant (1/s)Ka Parameter defined in Eq. (33)L Turbulent macroscale length (m)Mf Mass of feed (kg)Mr Mass of particle recovered (kg)n Flow behavior index (–)Nmb Number density of microbubble (1/m3)Np Number density of particles (1/m3)Qm Microbubble–particle mixture flow rate (m3/s)r Radius (m)Rc Percentage recovery (–)Rmax Maximum recovery (–)Uc Microbubble–particle circulation velocity (m/s)Up Particle velocity (m/s)

V2mb

� �1=2Root mean square velocity of microbubble (m/s)

V2p

� �1=2Root mean square velocity of particles (m/s)

t Time (s)Ti Induction time (s)Ts Sliding time (s)Z Collision frequency per unit volume (1/m3 s)

Greek letters

μe Effective viscosity of mixture (kg/m s)ν Kinematic viscosity (m2/s)ε Energy dissipation rate per unit mass (m2/s3)εg Gas holdup (–)ω Location of a particle at the gas–liquid interface (radian)ρf Density of fluid mixture (kg/m3)ρp Density of particle (kg/m3)θc Maximum collision angle (radian)θt Maximum possible collision angle defined in Eq. (34)θa Adhesion angle (radian)α Parameter defined in Eq. (3) (–)β Parameter defined in Eq. (35)σ Surface tension (N/m)τ Wall shear stress (Pa)γa Apparent shear rate (1/s)ΔP Pressure drop (N/m2)θ Contact angle (radian)

Dimensionless groups

Ca Capillary number (mfUc/σ) (–)Ren Non-Newtonian liquid Reynolds number Dn

c U2� nc ρ

8n� 1K4n

3nþ1

� �n� �(–)

References

Ahmadi, R., Khodadadi, D.A., Abdollahy, M., Fan, M., 2014. Nano-microbubble flo-tation of fine and ultrafine chalcopyrite particles. Int. J. Min. Sci. Technol. 24,559–566.

Ahmed, N., Jameson, G.J., 1985. The effect of bubble size on the rate of flotation offine particles. Int. J. Miner. Process. 14, 195–215.

Al-Shamrani, A.A., James, A., Xiao, H., 2002. Separation of oil from water by dis-solved air flotation. Colloids Surf. A Physicochem. Eng. Asp. 209, 15–26.

Alves, R.W., de Souza, A.A.U., Ulson de Souza, de Arruda Guelli, Jauregi, P., S.M.,2006. Recovery of norbixin from a raw extraction solution of annatto pigmentsusing colloidal gas aphrons (CGAs). Sep. Purif. Technol. 48, 208–213.

Amiri, M.C., Valsaraj, K.T., 2004. Effect of gas transfer on separation of whey proteinwith aphron flotation. Sep. Purif. Technol. 35, 161–167.

Bloom, F., Heindel, T.J., 1997. Mathematical modelling of the flotation deinkingprocess. Math. Comput. Model. 25, 13–58.

Chipfunhu, D., Zanin, M., Grano, S., 2011. The dependency of the critical contactangle for flotation on particle size – modelling the limits of fine particle flo-tation. Miner. Eng. 24, 50–57.

Cilliers, J.J., Bradshaw, D.J., 1996. The flotation of fine pyrite using colloidal gasaphrons. Miner. Eng. 9, 235–241.

Ciriello, S., Barnett, S.M., Deluise, F.J., 1982. Removal of heavy metals from aqueoussolutions using microgas dispersions. Sep. Sci. Technol. 17, 521–534.

Collins, G.L., Jameson, G.J., 1977. Double-layer effects in the flotation of fine parti-cles. Chem. Eng. Sci. 32, 239–246.

Couto, H.J.B., Nunes, D.G., Neumann, R., França, S.C.A., 2009. Micro-bubble sizedistribution measurements by laser diffraction technique. Miner. Eng. 22,330–335.

Dai, Z., Dukhin, S., Fornasiero, D., Ralston, J., 1998. The inertial hydrodynamicinteraction of particles and rising bubbles with mobile surfaces. J. ColloidInterface Sci. 197, 275–292.

Dai, Z., Fornasiero, D., Ralston, J., 1999. Particle–bubble attachment in mineral flo-tation. J. Colloid Interface Sci. 217, 70–76.

Dai, Z., Fornasiero, D., Ralston, J., 2000. Particle–bubble collision models – a review.Adv. Colloid Interface Sci. 85, 231–256.

Dobby, G.S., Finch, J.A., 1987. Particle size dependence in flotation derived from afundamental model of the capture process. Int. J. Miner. Process. 21, 241–260.

Duan, J., Fornasiero, D., Ralston, J., 2003. Calculation of the flotation rate constant ofchalcopyrite particles in an ore. Int. J. Miner. Process. 72, 227–237.

Dukhin, S.S., 1982. Role of inertial forces in flotation of small particles. Kolloid. Z. 44,431–441.

Fan, M., Tao, D., Honaker, R., Luo, Z., 2010. Nanobubble generation and its appli-cation in froth flotation (part I): nanobubble generation and its effects onproperties of microbubble and millimeter scale bubble solutions. Min. Sci.Technol. 20, 1–19.

Feng, W., Singhal, N., Swift, S., 2009. Drainage mechanism of microbubble disper-sion and factors influencing its stability. J. Colloid Interface Sci. 337, 548–554.

Finch, J.A., Dobby, G.S., 1990. Column. flotation. Pergamon Press, Oxford.Fuda, E., Jauregi, P., 2006. An insight into the mechanism of protein separation by

colloidal gas aphrons (CGA) generated from ionic surfactants. J. Chromatogr. B843, 317–326.

Gaudin, A.M., Mular, A.L., O'Connor, R.F., 1960. Separation of Microorganisms byFlotation: II. Flotation of spores Bacillus subtilis Var. Niger. Appl. Microbiol. 8,91–97.

Han, M., Dockko, S., 1998. Zeta potential measurement of bubbles in DAF processand its effect on the removal efficiency. KSCE J. Civ. Eng. 2, 461–466.

Hanotu, J., Bandulasena, H.C.H., Zimmerman, W.B., 2012. Microflotation perfor-mance for algal separation. Biotechnol. Bioeng. 109, 1663–1673.

Jameson, G.J., Nam, S., Moo Young, M., 1977. Physical factors affecting recovery ratesin flotation. Miner. Sci. Eng. 9, 103–118.

Jarudilokkul, S., Rungphetcharat, K., Boonamnuayvitaya, V., 2004. Protein separa-tion by colloidal gas aphrons using nonionic surfactant. Sep. Purif. Technol. 35,23–29.

Jauregi, P., Gilmour, S., Varley, J., 1997. Characterisation of colloidal gas aphrons forsubsequent use for protein recovery. Chem. Eng. J. Biochem. Eng. J. 65, 1–11.

Johnson, D., Hilal, N., Waters, K., Hadler, K., Cilliers, J., 2009. Measurements ofInteractions between particles and charged microbubbles using a combinedmicro- and macroscopic strategy. Langmuir 25, 4880–4885.

Krasowska, M., Malysa, K., 2007. Kinetics of bubble collision and attachment tohydrophobic solids: I. Effect of surface roughness. Int. J. Miner. Process. 81,205–216.

Larmignat, S., Vanderpool, D., Lai, H.K., Pilon, L., 2008. Rheology of colloidal gasaphrons (microfoams). Colloids Surf A Physicochem. Eng. Asp. 322, 199–210.

Li, D., Fitzpatrick, J.A., Slattery, J.C., 1990. Rate of collection of particles by flotation.Ind. Eng. Chem. Res. 29, 955–967.

McDonald, B.T., Cui, T., 2011. Superhydrophilic surface modification of copper sur-faces by layer-by-layer self-assembly and liquid phase deposition of TiO2 thinfilm. J. Colloid Interface Sci. 354 (1), 1–6.

Matis, K.A., Mavros, P., 1991. Recovery of metals by ion flotation from dilute aqu-eous solutions. Sep. Purif. Rev. 20, 1–48.

Noble, M., Brown, A., Jauregi, P., Kaul, A., Varley, J., 1998. Protein recovery using gas–liquid dispersions. J. Chromatogr. B Biomedical Sci. Appl. 711, 31–43.

Parmar, R., Majumder, S.K., 2013. Microbubble generation and microbubble-aidedtransport process intensification-A state-of-the-art report. Chem. Eng. Process.64, 79–97.

R. Parmar, S.K. Majumder / Chemical Engineering Science 142 (2016) 42–5454

Parmar, R., Majumder, S.K., 2014. Hydrodynamics of microbubble suspension flowin pipes. Ind. Eng. Chem. Res. 53, 3689–3701.

Parmar, R., Majumder, S.K., 2015. Terminal rise velocity, size distribution and sta-bility of microbubble suspension. Asia-Pac. J. Chem. Eng. 10, 450–465.

Pope, S.B., 2000. Turbulent Flows. Cambridge University Press, Cambridge, UK.Poljacek, S.M., Cigula, T., Gojo, M., 2008. Formation and defining the different

aluminium oxide microstructures in alkaline solutions. Int. J. Mater. Form. 1 (1),463–466.

Pyke, B., Fornasiero, D., Ralston, J., 2003. Bubble particle heterocoagulation underturbulent conditions. J. Colloid Interface Sci. 265, 141–151.

Ralston, J., 1983. Thin films and froth flotation. Adv. Colloid Interface Sci. 19, 1–26.Ralston, J., Dukhin, S.S., Mishchuk, N.A., 2002. Wetting film stability and flotation

kinetics. Adv. Colloid Interface Sci. 95, 145–236.Reay, D., Ratcliff, G.A., 1973. Removal of fine particles from water by dispersed air

flotation: effects of bubble size and particle size on collection efficiency. Can. J.Chem. Eng. 51, 178–185.

Rousseau, R.W., 1987. Handbook of Separation Process Technology. Wiley, NewYork.

Roy, D., Valsaraj, K.T., Kottai, S.A., 1992. Separation of organic dyes fromwastewaterby using colloidal gas aphrons. Sep. Sci. Technol. 27, 573–588.

Rulison, C., 2000. Two-Component Surface Energy Characterization as a Predictor ofWettability and Dispersability. AN213/CR, KRÜSS Application Note-213.

Saulnier, P., Lachaise, J., Morel, G., Graciaa, A., 1996. Zeta potential of air bubbles insurfactant solutions. J. Colloid Interface Sci. 182, 395–399.

Schubert, H., 1999. On the turbulence-controlled microprocesses in flotationmachines. Int. J. Miner. Process. 56, 257–276.

Schulze, H.J., 2003. Flotation as a heterocoagulation process: possibilities of cal-culating the probability of flotation. In: Dobias, B. (Ed.), Surfactant ScienceSeries (Coagulation and Flocculation). Marcel Dekker, New York, pp. 321–354.

Shah, D.O., Djabbarah, N.F., Wasan, D.T., 1978. A correlation of foam stability withsurface shear viscosity and area per molecule in mixed surfactant systems.Colloid Polym. Sci. 256, 1002–1008.

Shen, M., Wheelock, T.D., 2000. Coal agglomeration with microbubbles. Coal Prep.21, 277–298.

Shen, Y., Longo, M.L., Powell, R.L., 2008. Stability and rheological behavior of con-centrated monodisperse food emulsifier coated microbubble suspensions. J.Colloid Interface Sci. 327, 204–210.

Shibata, E., Saito, S., Nakamura, T., 2008. Froth separation of ferrihydrite slurry usingmicrobubbles with ultrasonic irradiation. Materials Transactions 49 (7),1681–1687.

Solari, J.A., Gochin, R.J., 1992. Fundamental aspects of microbubbles flotation. In:Ralston, J., Laskowski, J.S. (Eds.), Colloid Chemistry in Mineral Processing, vol.12. Development in Mineral Processing, Amsterdam, pp. 395–418.

Sutherland, K.L., 1948. Physical chemistry of flotation. XI. Kinetics of the flotationprocess. J. Phys. Colloid Chem. 52, 394–425.

Subedi, D.P., Madhup, D.K., Joshi, U.M., Sharma, A., Huczko, A., 2011. Study of thewettability of ZnO nanofilms. International. Nano Lett. 1 (2), 117–122.

Sven-Nilsson, I., 1935. Effect of contact time between mineral and air bubbles onflotation. Kolloid-Z. 69, 230–232.

Takahashi, M., 2005. ζ potential of microbubbles in aqueous solutions: electricalproperties of the gas�water interface. J. Phys. Chem. B 109, 21858–21864.

Trahar, W.J., Warren, L.J., 1976. The flotability of very fine particles – a review. Int. J.Miner. Process. 3, 103–131.

Tseng, H., Pilon, L., Warrier, G.R., 2006. Rheology and convective heat transfer ofcolloidal gas aphrons in horizontal mini-channels. Int. J. Heat. Fluid Flow. 27,298–310.

Tuteja, R.K., Spottiswood, D.J., Misra, V.N., 1994. Mathematical models of the col-umn flotation process a review. Miner. Eng. 7, 1459–1472.

Waters, K.E., Hadler, K., Cilliers, J.J., 2008. The flotation of fine particles usingcharged microbubbles. Miner. Eng. 21, 918–923.

Weber, M.E., Paddock, D., 1983. Interceptional and gravitational collision effi-ciencies for single collectors at intermediate Reynolds numbers. J. ColloidInterface Sci. 94, 328–335.

Woo, S.W., 1971. Simultaneous Free and Forced Convection Around SubmergedCylinders and Spheres (PhD. thesis). McMaster University, Hamilton, Ontario.

Wu, H., Patterson, G.K., 1989. Laser-doppler measurements of turbulent-flowparameters in a stirred mixer. Chem. Eng. Sci. 44, 2207–2221.

Xu, Q., Nakajima, M., Ichikawa, S., Nakamura, N., Roy, P., Okadome, H., Shiina, T.,2009. Effects of surfactant and electrolyte concentrations on bubble formationand stabilization. J. Colloid Interface Sci. 332, 208–214.

Ye, Y., Khandrika, S.M., Miller, J.D., 1989. Induction-time measurements at a particlebed. Int. J. Miner. Process. 25, 221–240.

Yi-jun, C., Lin-ping, S., Xin-hua, Y., 2009. Research on 730 series frother for coppersulfide ore separation by cyclonic-static microbubble flotation column. Proce-dia Earth Planet. Sci. 1, 771–775.

Yoon, R.-H., Yordan, J.L., 1986. Zeta-potential measurements on microbubblesgenerated using various surfactants. J. Colloid Interface Sci. 113, 430–438.

Yoon, R.-H., Yordan, J.L., 1991. Induction time measurements for the quartz—amineflotation system. J. Colloid Interface Sci. 141, 374–383.