Harvesting and dewatering yeast by microflotation

Biochemical Engineering Journal 82 (2014) 174–182

Contents lists available at ScienceDirect

Biochemical Engineering Journal journal homepage: www.elsevier.com/locate/bej

Regular Article

Harvesting and dewatering yeast by microflotation James Hanotu a,∗ , Esther Karunakaran a , Hemaka Bandulasena b , Catherine Biggs a , William B. Zimmerman a a b

Department of Chemical and Biological Engineering, University of Sheffield, Mappin Street, Sheffield S1 3JD, United Kingdom Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, United Kingdom

a r t i c l e

i n f o

Article history: Received 15 May 2013 Received in revised form 16 October 2013 Accepted 23 October 2013 Available online 6 November 2013 Keywords: Bioflocculant Chitosan Dispersed air flotation Dissolved air flotation Fluidic oscillation Microflotation

a b s t r a c t Microbubble has been applied for the recovery of yeast cells from their growth medium using the bioflocculant–chitosan. Results reaching 99% cell recovery were obtained under various conditions examined. The result of bubble size distribution showed that mean bubble size increased as microbubble diffuser pore size was increased. Also, cell recovery efficiency was a function of both bubble size and particle size (cell size). For smaller particles (<50 ␮m), relatively smaller bubbles (<80 ␮m) were found to be more effective for recovery, otherwise, relatively larger bubbles (80–150 ␮m) proved to be efficient in recovering larger particles (particle size: ∼250 ␮m). Acidic and neutral pHs were effective in separation as hydrophobic particles were formed. As pH tends toward alkalinity, flocs become more hydrophilic, leading to low recovery from the aqueous solution. In addition, separation efficiency was dependent on flocculant dose as increase in concentration improved flocculation and consequently, yeast recovery. However, above a critical concentration, overdosing occurred and inadvertently, recovery efficiency decreased. The application of chitosan as a bioflocculant and the subsequent application of microflotation for the separation of yeast cells proved effective and promises several advantages over non-bubble based separation techniques that preclude continuous industrial-scale production. © 2013 Published by Elsevier B.V.

1. Introduction Yeast is one of the most well studied microorganisms and has been long used as the main ingredient in many food products [7]. Their application in the production of bioethanol, as solvent for pharmaceutics, a renewable source of energy [21], feedstock in alcoholic beverage production [7] or animal feed has also become widespread. Bioremediation of wastewaters containing heavy metals can also be achieved using yeast cells, owing to their ability to remove a wide range of metals [15] with fast separation of biomass upon treatment [15]. It has also been exploited for the production of heterologous products such as vaccines and human hormones [4] and they offer apart from their abundance, a cost effective option [23]. One of the most important stages in the utilization of yeast in any of the above operations is the harvesting and dewatering of the yeast biomass. Yeast must be harvested from culture medium for further processing. Several methods have been designed and developed for the recovery of yeast from culture medium. However, their small size and relatively high cell density pose numerous separation difficulties. Some of the traditional techniques employed such

∗ Corresponding author. Tel.: +44 7529572107. E-mail address: [email protected] (J. Hanotu). 1369-703X/$ – see front matter © 2013 Published by Elsevier B.V. http://dx.doi.org/10.1016/j.bej.2013.10.019

as filtration and sedimentation are time consuming and relatively inefficient, preventing their continuous large-scale application in industries [12]. Given these challenges, yeast harvesting and dewatering has become a significant concern across several overlapping industries and could contribute substantially to the total production cost with potential impacts on food prices. More efficient techniques must be sought to overcome the challenge with yeast separation from culture medium. Bubble based separation techniques provide an advantageous approach to harvesting yeast because flotation is a rate intensifying separation process by enhancing buoyancy force over sedimentation (Molina et al. [17]). Dissolved air flotation (DAF) and Jameson’s cells are industry’s most employed bubble based separation techniques [6] but yet untried in yeast cell recovery from medium. Their application however, is likely unfeasible largely due to their intrusive nature but also the high energetic consumption associated with their operations. Another main reason for their unsuitability is the high shear from the exit nozzle due to the high operating conditions (pressure ∼6 bar) [6,9]. The key challenge is to change from high power consumption regimes to low power consumption methodologies without sacrificing performance and the yeast biomass quality. Here, we address all these concerns with microflotation [9] that applies fluidic oscillation induced microbubble clouds [26], already applied for gas exchange in accelerating algal biomass growth [27] and

J. Hanotu et al. / Biochemical Engineering Journal 82 (2014) 174–182

175

in microalgae harvesting and dewatering. Microflotation is one approach that achieves low power consumption with desirable bubble size and flux under laminar flow regime. In DAF systems the energetic release of supersaturated liquid in the flotation cell for microbubble generation results in high turbulent flow and inadvertently floc break-up. This counter-productive behavior is not experienced in microflotation, as only gas is injected to produce bubble under atmospheric pressure. Thus the study aims to investigate the application and performance of fluidic oscillator generated microbubbles in yeast separation using standard methodologies for flotation separations as well as the effect of chitosan on recovery efficiency. In addition, the study explores the effect of varying bubble sizes on flotation of particles. This paper is organized as follows: In Section 2, the materials and methods are presented followed by the results and discussions in Section 3. 2. Materials and methods 2.1. Material preparation Sterile Yeast Peptone Dextrose (YPD) medium was made using yeast broth and yeast extract (Sigma Aldrich, UK). 8.5 g of the yeast broth and yeast extract respectively were added to 1 L distilled water and mixed until dissolved before sterilizing with high pressure saturated steam for about 15 min. Meanwhile, chitosan (Sigma Aldrich, UK) stock was made by dissolving 5 g of dry chitosan (Sigma Aldrich, UK) in 150 mL 0.5 M HCl (Sigma Aldrich, UK) which gives a viscosity of 0.9 Pa s. 2.2. Experimental procedure After pH adjustment of the growth medium, 1 g of dried yeast (Saccharomyces cereviseae, Lallemand, UK) was reconstituted into 1 L of growth medium and mixed for 1 min to form a homogenous dispersion before chitosan was added. Rapid coagulation with a motorized stirrer at 3500 rpm followed for 1 min before the mixture was stirred for a further 1 min under low speed at 75 rpm to promote floc growth. After flocculation, the microbubble generator was turned on and the mixture was gradually introduced into the flotation rig where cells were harvested for 20 min. Samples were collected every 2 min for optical density measurements. Biomass concentration correlates with optical density (OD) and was measured by spectrophotometer DR 2800 (Hach Lange, UK) to determine OD at 660 nm. The experimental rig is shown in Fig. 1. The rig consists of a fluidic oscillator, a microbubble diffuser and a microflotation column. Bubbles were generated by fluidic oscillation [9,25]. For each run, the microbubble diffuser was fitted with different membranes with pore size: 25, 50, 75, 100, 125 ␮m respectively. Also, the chitosan concentration was varied (0.2; 0.4; 0.6; 0.8; 1%, v/v) as was the pH of the growth medium (pH 5, 7 and 9). All experiments were conducted under room temperature (21 ◦ C). Recovery efficiency was determined using the equation below:

Fig. 1. Schematic representation of the experimental set-up. Compressed air (0.8 bars) is fed into the oscillator, which then feeds the microbubble diffuser with a portion of the air bleed-off or channeled otherwise to another set of diffuser. In this study, a portion of the air was bled off downstream of the fluidic oscillator.

2.4. Zeta potential measurement of yeast cells Dry yeast was reconstituted in 10 mL YPD medium and placed in an incubator at 30 ◦ C and 300 rpm to keep cells from settling. Following that, 3.7 g agar was made up in 250 mL YPD medium to a concentration of 1.5% (w/v) before pouring into plates and allowed to set. Then, 20 ␮L of the reconstituted yeast was cultured in the YPD agar medium for 24 h. Zeta potential was measured with the zeta potential analyzer (Brookhaven ZetaPALS, UK) using the phase amplitude light-scattering method. Samples were centrifuged at 3000 × g for 5 min. After which cells were washed and re-suspended twice in 100 mM potassium chloride (KCl) before centrifugation at 3000 × g for 5 min. An electric field of ∼2.5 V/cm was used during zeta potential measurements [18]. Triplicate measurements of samples of cells were done for reproducibility. Results represent the average of ten successive runs. 2.5. Yeast particle size measurement

where Ci and Cf are the initial and final yeast concentrations at mid-depth respectively.

The particle size distribution of the yeast cells/flocs was measured with the Mastersizer S (Malvern Instrument, UK). Cells were measured with and without the addition of a chitosan. Yeast cells (1 g) were reconstituted in growth medium and immediately dispersed by stirring for 1 min. Under no coagulant conditions, cells were measured immediately. Otherwise, chitosan was added at varying concentrations (0.2; 0.4; 0.6; 0.8; 1%, v/v) and the mixture rapidly mixed (coagulation) for 1 min. Next the sample was gently added to the Mastersizer until an obscurity of 15–20% was attained with the dispersion unit stirring at 1200 rpm. The stirring rate was chosen in order not to cause floc breakage but also to facilitate good dispersion around the measuring device. Samples were taken for measurement soon after rapid mixing to avoid floc settling.

2.3. Bubble size measurement

3. Results and discussion

The measurement of the size distribution of gas bubbles was carried out by high-speed photography and image analysis according to the method described by Hanotu et al. [10,11].

3.1. Bubble size measurements

R=

Ci − Cf Ci

× 100.

Bubble size was measured in water as well as in the yeast culture medium with the mean bubble size, a function of the diffuser


160 Water

400

140

Culture Medium

120 300

100 80

200

60 40

100

20

-26 -24 Zeta Potential (mV)

Mean Bubble Diameter in Water m)

500

Mean Bubble Diameter in Yeast Medium ( m)

176

0

0 25

0

50 75 100 Mesh Pore Size (um)

125

-22 -20 -18 -16

150

-14 3

Fig. 2. Mean bubble size versus diffuser membrane pore size. The error bars represent standard error.

4

5

6

7

pH

8

9

10

11

12

Fig. 3. Zeta potential analyses of yeast cells at varying pHs after reconstituting in growth medium at time t(0).

membrane pore shown in Fig. 2. In the yeast medium, a mean bubble size of 61 ␮m was recorded from the 25 ␮m pore size membrane and for the 50 and 75 ␮m membrane mean bubble size of 68 ␮m and 87 ␮m respectively was recorded. As membrane pore size was further increased, higher average sizes (104 ␮m and 140 ␮m) were observed for the 100 and 125 ␮m pore size. For bubbles generated in the growth medium, the bubbles are almost same size as their exit pores. Bubbles produced in water are however approximately

Zeta potential of yeast cells is dependent on strain type, culture condition and growth phase. The iso-electric point is likely

0% Chitosan 0.2% v/v Chitosan 0.4% v/v Chitosan 0.6% v/v Chitosan 0.8% v/v Chitosan 1% v/v Chitosan 1.2% v/v Chitosan

10

% Volume

% Volume

6

3.2. Zeta potential measurements

(b) 12

(a) 10 8

2–3 times larger than their exit pores. The difference in size is attributed to the surface tension (wettability) of both liquids.

4

8 6

0% Chitosan 0.2% v/v Chitosan 0.4% v/v Chitosan 0.6% v/v Chitosan 0.8% v/v Chitosan 1% v/v Chitosan 1.2% v/v Chitosan

4 2 0 0.01

2 0 0.01

1 100 Particle Size (µm)

1 Particle Size (µm)

100

Mean Particle Size (µm)

(c) 300 pH 5

250

pH 7

200 150 100 50 0 0

0.5 1 Chitosan Concentration (%v/v)

1.5

Fig. 4. Size distribution of yeast cells at varying flocculant concentratons and pH for yeast floc sizes. (a) Size distribution at pH 5; (b) size distribution at pH 7. (c) Combined plot of mean particle size.

J. Hanotu et al. / Biochemical Engineering Journal 82 (2014) 174–182 Table 1 Zeta potential values ± standard deviation. Zeta potential values calculated from electrophoretic mobility using Smoluchowski’s formula. 5

pH of growth pH during measurement

Zeta potential (mV): 7 h post inoculation 5 −12.0 ± 7 −12.1 ± 9 −13.6 ± Zeta potential (mV): 17 h post inoculation 5 −10.7 ± 7 −13.8 ± 9 −15.1 ±

7

9

1.4 1.3 3.1

−8.1 ± 1.5 −13.8 ± 2.6 −14.7 ± 2.7

−7.2 ± 2.6 −13.8 ± 2.9 −11.8 ± 1.5

1.0 1.7 0.9

−13.7 ± 1.3 −15.3 ± 2.1 −15.3 ± 0.8

−12.8 ± 2.5 −16.4 ± 0.8 −17.2 ± 1.4

to be similarly influenced. However, provided a sufficient concentration of flocculant is dosed to reduce the magnitude of the cell surface charge (i.e., charge approaching 0 – isoelectric point), agglomeration of cells will inevitably occur under the right pH and consequently bubble–particle attachment probability will increase. The results of the zeta potential measurements are shown in Fig. 3. Samples were measured immediately after reconstituting cells in growth medium. In all the conditions analyzed, the isoelectric point (pH or zero charge) of the cells was not reached indicating that the cell surface is composed primarily of anionic moieties such as phosphates and carboxylates found in polysaccharides [20]. Since information on the isoelectric point was not obtained, subsequent conditions and replicates were assayed only at the relevant pH, i.e. 5, 7 and 9. The results are provided in Table 1. The zeta potential values in Table 1 are an average of two biological replicates ± standard deviation. However, no significant change was recorded in the zeta potential values across the different culture growth phases. This probably indicates that the cell growth phase is unlikely to be the rate-limiting factor in separation.

3.3. Particle size distribution The average particle size (yeast cell) for each flocculant dose is plotted in Fig. 4 for both pH 5 and pH 7. Size of particles increased as particles aggregated owing to the addition of chitosan. The size of the yeast cells (without flocculant) was found to range from 17 to 20 ␮m but otherwise, particle size reached 251 ␮m. Average particle size increased with flocculant concentration until a critical flocculant concentration was exceeded. Across the flocculant concentrations investigated, flocs generated at pH 5 (Fig. 4a) showed a relatively narrow size distribution. Contrastingly, flocs generated under pH 7 (Fig. 4b) exhibited a wider size range and are larger than flocs produced under pH 5. The size distribution results agree well with recovery efficiency results under the same conditions (Figs. 5–7). 3.4. Effect of chitosan dose The influence of chitosan on the performance of yeast separation was investigated by varying the concentration of chitosan in the medium. The results are presented in Figs. 5 and 6. At pH 5, recovery efficiency improved with increase in chitosan concentration from 9% at 0.2% (v/v) to ∼99% at 0.4% and 0.6% (v/v) similarly before slightly dropping at 0.8% (v/v) to 98%. Further decrease in efficiency to ∼88% was recorded at increased chitosan concentration of 1% (v/v). Overall across the chitosan range investigated, separation efficiency increased with increasing chitosan dosage up to a maximum before dropping with further dosage. However, the recovery efficiency showed a sharp increase with time before reaching a plateau. At optimum concentrations (0.6–0.8%, v/v), the plateau was achieved within 1–2 min after start of separation but otherwise ∼10 min later. The success of flotation separation is largely dependent on the mechanisms governing particle–bubble interaction. Separation or collection efficiency as proposed by Derajguin and Dukhin [28] is a product of three sub-steps viz.; particle–bubble collision,

(b)100 Recovery Efficiency (%)

Recovery Efficiency (%)

(a) 100 0.2 %v/v

80

25 m PSM 50 m PSM 75 m PSM 100 m PSM 125 m PSM

60 40 20 0

80 60 0.4% v/v 25 m PSM

40

50 m PSM 75 m PSM

20

100 m PSM 125 m PSM

0 0

5

10

15

20

25

0

5

10

Time(mins)

60 0.6% v/v

40


0

80 60 0.8% v/v

40


20

5

10 15 Time(mins)

20

25

25

80 60 1% v/v

40


20 0

0 0

20

(e) 100 Recovery Efficiency (%)

80

15

Time(mins)

(d) 100 Recovery Efficiency (%)


(c) 100

20

177

0

5

10 15 Time(mins)

20

25

0

5

10

15

20

25

Time(mins)

Fig. 5. Plots of recovery efficiency with time at pH 5 under varying membrane pore sizes and chitosan concentrations of: (a) 0.2% (v/v); (b) 0.4% (v/v); (c) 0.6% (v/v); (d) 0.8% (v/v); (e) 1% (v/v). N.B: PSM – pore size of membrane.


(b) 0.2% v/v

80


60

100

(c) 100

80

80


40 20 0

60

0.4% v/v

40

25 m PSM 50 m PSM 75 m PSM 100 m PSM

20

125 m PSM

10 15 Time(mins)

20

25

10 20 Time(mins)

0

30

(d) 100

(e) 100

80

80


5

60

0.8% v/v 25 m PSM 50 m PSM 75 m PSM 100 m PSM 125 m PSM

40

20

0 0

5

10

60

0.6% v/v

40


20 0

0 0



(a) 100


178

15

20

Time(mins)

25

0

5

10 15 Time(mins)

60

1% v/v

40


20

25

20

0 0

5

10 15 Time(mins)

20

25

Fig. 6. Plots of recovery efficiency with time at pH 7 under varying membrane pore sizes and chitosan concentrations of: (a) 0.2% (v/v); (b) 0.4% (v/v); (c) 0.6% (v/v); (d) 0.8% (v/v); (e) 1% (v/v). N.B: PSM – pore size of membrane.

attachment and the aggregate stability efficiencies. The product of these three processes must be unity for optimum collection efficiency. In practice, collision and attachment efficiencies are connected through the drainage and rupture of the thin liquid film separating the particle and bubble but are however independent steps that are usually considered separately. Whilst collision efficiency (Ec ) is a function of the ratio of particle size to bubble size [24] and thus relatively low for small particles and coarse bubbles, attachment efficiency (Ea ) is mainly influenced by particle zeta potential and floc size [3,13,24]. Nonetheless, stability efficiency (Es ) is largely dependent on inertial force and system hydrodynamics in the flotation cell [24]. Given conditions when collision efficiency is unity, attachment and stability efficiencies become the rate limiting factors. Therefore, increase in chitosan concentration, increases particle hydrophobicity and particle zeta potential shifts further toward the iso-electric point, promoting


120 100 80 60 40 pH 5 20

pH 7

0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

Chitosan Concentration (% v/v) Fig. 7. Effect of chitosan dose on yeast cell recovery efficiency after 20 min across pH 5 and 7. Recovery efficiency increased with chitosan dose up to an optimum concentration. The error bar represents standard error.

particle agglomeration. As a consequence, attachment and stability efficiencies increase as seen in Fig. 5. But at low concentrations however, attachment and stability efficiencies become less than unity and resultantly, recovery efficiency drops. Furthermore, when the repulsive forces that exist due to the presence of the double layer are high, colloidal particles will repel each other and hence prohibit agglomeration. In such an instance, the particle must be destabilized by pre-treatment through one of the four known destabilization mechanisms for colloids. From Fig. 3 it can be observed that the particle (yeast cells) iso-electric point was not attained across pH 4–10. The energy forces between colloidal particles must be balanced for optimum agglomeration of particles. In the presence of enough counter ions, colloidal particles become electrically neutral (Isoelectric point). Under this state, optimum flocculation can be expected. Contrarily, insufficient flocculant dose often leads to partial particle destabilization and ultimately poor floc formation. On occasion, some particles still remain completely stabilized in the liquid medium. This explains the low recovery efficiency obtained at chitosan concentration dose of 0.2% (v/v) (see Figs. 5a and 6a). By converse, over-dosing with chitosan is similarly counter productive. To investigate this hypothesis, tests were conducted under pH 5 and 7 with incremental chitosan addition until efficiency began to drop (see Fig. 7). Excessive chitosan dosage re-stabilizes particles. Cheng et al. [2] reported a similar observation and suggested particle re-suspension was due to the reversal of surface charge at higher doses. Similarly, results at neutral pH showed increase in harvesting efficiency as chitosan dosage increased. However, pH 7 revealed a rather higher tolerance for flocculant concentration (Fig. 7). This outcome corroborates the findings of Cheng et al. [2] with recovery of organic matters from brewery wastewater. Divakaran and Pillai [5] also reported that particles exhibited a higher tolerance for chitosan flocculation at pH 7 as well as optimum algal removal efficiency under the same pH state.


179

Fig. 8. SEM photomicrograph of yeast cells. (a) Reconstituted yeast cells. (b) yeast cells with the bioflocculant–chitosan at pH 7.

3.5. Effect of pH

3.6. Effect of bubble size

As with other coagulant and flocculants, the performance of chitosan as a bioflocculant is highly influenced by medium pH (Figs. 5 and 6). Acidic and neutral pH conditions of the culture medium favored cell harvest more than alkaline states. Recovery efficiency results reaching 99% was obtained under both acidic and neutral pH states. By converse however, no significant separation was recorded as pH became alkaline (Fig. 10c: result from preliminary studies for the optimum operating conditions). The pH effect can be attributed to the protonation difference of the amine groups and changes in the macromolecular chain conformation of chitosan. pH affects not only the size but the structure of flocs (see Fig. 8). In neutral pH, chitosan due to its coiled structure generates larger flocs (Fig. 4). Conversely under acidic state, the biopolymer has increased charge density and extended chain and as such generates relatively smaller, largely less dense flocs [14]. Given their hydrophobic nature and unstable intervening thin liquid film, the flocs generated are readily susceptible to aggregation with microbubbles. Additionally, Hewitt et al. [13] experimentally showed that attachment efficiency (Ea ) increases with increasing particle hydrophobicity. A completely different outcome however, was observed under pH 9 where no significant yeast cell recovery was recorded across the different chitosan doses. Since no significant results were recorded, the results have not been included here. However, a preliminary result is provided in Fig. 10, showing comparison with sedimentation. One reason for this outcome is the gelatinous flocs formed. Gelatinous flocs have been widely reported at higher pH levels [8,9]. Generally, gelatinous flocs often have slight negative charge and high affinity for the containing medium (aqueous phase). Apart from being hydrophilic, gelatinous flocs have slippery surfaces and the intervening liquid sheet existing between a particle and bubble is usually stable with hydrophilic surfaces [16], leading to no liquid film drainage and reduced attachment efficiency and consequently decreased collection efficiency. Even in rare instances when collision occurs, hydrophilic particles do not adhere to the surface of air bubbles [24]. Gochin and Solari [8] using dissolved air flotation (DAF) also reported that hydrophilic quartz particles or flocs would not be recovered. Their dispersion is stabilized by hydration and as such are thermodynamically stable. Agglomeration of hydrophilic colloids requires the significant dosage of ions, which compete for water molecules with the colloids, thereby causing dehydration of the colloidal particles.

The mean bubble size result presented in Fig. 2 showed that average bubble size was influenced by the liquid type and varied directly proportionate to diffuser pore size. The effect of bubble size on recovery efficiency was investigated and the result presented in Fig. 9. From the results, recovery efficiency was influenced by bubble size but also important was effect of the particle size. It might seem that under the influence of the high bubble density formation the underlying principles of effects of bubble–particle induction time can be neglected. The experimental results above suggest otherwise. Particle/floc size is essential in determining the average bubble size required for a flotation [24]. But it is dependent on flocculant concentration as is the recovery efficiency. Note how for a given floc size (Fig. 9), efficiency generally increases with bubble size to a maximum before gradually dropping with further increase in bubble size. One main justification for this outcome is the low terminal rise velocity of microbubbles, which is intrinsically linked to their low buoyancy as well as size but also influential, is the changing floc size at varying flocculant concentration. At reduced chitosan concentration (Fig. 9a), small flocs (see Fig. 4) are produced due to insufficient counter ions necessary for particle destabilization. Particle aggregation is largely low given this condition, yielding relatively small, loose and less dense flocs. Thus, recovery efficiency favors smaller microbubbles (<70 ␮m) because smaller microbubbles are gentler with small, loose and less dense flocs for their collision kinetic energy is too small to distort or break the flocs. Also, the probability of particle–bubble collision is a function of the ratio of particle to bubble size and varies indirectly with bubble size for small particles [24]. Under this condition where particle density approaches density of surrounding fluid, long-range hydrodynamic interaction (LRHI) influence dominates particle–bubble collision mechanism and therefore dictates the trajectory of the particles with respect to the fluid streamlines [16]. However, as the (particles) flocs increase in size and become denser with higher flocculant dose (see Fig. 9b graph: 0.4%, v/v), the optimum recovery efficiency shifts toward relatively larger microbubbles (∼70–90 ␮m). Inertial forces become the influential collision mechanism given the inability of coarse and dense flocs to follow fluid streamlines and also given that their densities are greater than the containing medium, they posses a settling velocity which deviates their trajectory from the fluid streamlines [16]. Microbubbles experience a tangential stress due to this settling velocity [19]. Furthermore, this stress decelerates bubbles,

180


ultimately causing their terminal rise velocity to approach that of a solid sphere and as consequence resulting in an increase in particle–bubble residence time. This is a key reason for the low efficiency below the bubble size range 70–90 ␮m. Above this range, however, the recovery efficiency drops as the size ratio of particle to bubbles becomes low due to the increased bubble size. Boussinesq [1] first described the tangential stress effect on microbubble as ‘surface viscosity’. Velocity results less than those predicted by Stokes’ law were reported by Takahashi [22] when the author studied the rise velocity of bubble swarm (10–55 ␮m) produced by a vortex in distilled water. Auspiciously, as microbubble size increases, their buoyant force increases consequently and balances out this tangential stress. In other words, larger bubbles experience less tangential stress for a given particle size. Therefore, further increase in flocculant concentration (Fig. 9c: 0.6 and 0.8%, v/v) results in much coarser and denser flocs and again, efficiency is observed to shift favorably toward relatively large bubbles (90–100 ␮m). A limit is reached nonetheless, where further flocculant addition yields no more increase in floc size. Beyond this limit actually (Fig. 9d: 1%, v/v), overdosing occurs due to excess flocculant concentration and resultantly, floc size significantly reduces. Therefore, optimum efficiency tips back toward smaller microbubbles (70–90 ␮m) again. Excessive flocculant dosage contributes to particle re-suspension and reduction in process efficiency [14]. Note however, that whilst the critical coagulant concentration (CCC) is reached at 8% (v/v) for pH 5, pH 7 shows a higher tolerance for flocculant dose (see Fig. 7).

3.7. Effect of harvest method In order to explore the effectiveness of microflotation against a control, tests were conducted with and without bubbles to simulate flotation and sedimentation separation techniques respectively. Fig. 10 displays the optimal results for either technique under the three pH states. Yeast recovery by microflotation and sedimentation was 98% and 93% at pH 5 and 99% and 93% at pH 7 respectively but at pH 9, microflotation yielded no significant separation whilst sedimentation resulted in 72% recovery efficiency. The results from pHs 5 and 7 are comparative. Higher recovery efficiency was obtained with microflotation than sedimentation to the tune of 6%. Under pH 9, no significant separation was recorded (with the introduction of microbubbles) due to the gelatinous hydrophilic flocs formed. Collision and attachment probability is low with hydrated colloids. Therefore, recovery by sedimentation yielded higher efficiency. It is worth noting however, the overall drop in recovery efficiency under alkaline pH. The substantial recovery efficiency obtained by sedimentation can be explained by the high cell density and large size of yeast cells (17–20 ␮m), thus facilitating rapid sedimentation. In addition, the calculation of zeta potentials using the Smoluchowski approximation (Fig. 3), suggests that the zeta-potential of the cell surface ranged approximately between −8 and −23 mV across the various culture conditions suggesting that the yeast suspensions in 100 mM KCl were not stable and likely to sediment over time.

(b) 99 0.2 %v/v

50



(a) 60

40 30 20

0.4 %v/v

98 97 96 95 94

10

93

0 0

25

50 75 100 125 Mesh Pore Size ( m)

0

150

(c) 99

25

50 75 100 125 Mesh Pore Size ( m)

150

(d) 94

97 96

0.6 %v/v 0.8 %v/v

95 94



1 %v/v 98

92 90 88 86 84

93 0

25

50 75 100 125 Mesh Pore Size ( m)

150

0

25

50 75 100 125 Mesh Pore Size ( m)

150

Fig. 9. Plots of recovery efficiency against bubble size and varying flocculant concentrations: (a) 0.2% (v/v); (b) 0.4% (v/v); (c) 0.6% (v/v), 0.8% (v/v), 1% (v/v).


80 60 Microflotation Sedimentation

40 20

80 60 40

Microflotation Sedimentation

20 0

0 0

5

10

15

20

25

0

5

Time(mins)

10

15

20

25




(c) 100

(b) 100

(a) 100

181

80 60 40

Microflotation Sedimentation

20 0 0

Time(mins)

5

10

15

20

25

Time(mins)

Fig. 10. Comparison of recovery efficiencies with time of yeast cells between two separation techniques: microflotation and sedimentation at varying pH conditions. (a) pH 5; (b) pH 7; (c) pH 9.

Although there was slight improvement in separation efficiency with microbubbles over sedimentation, this adds up to substantial savings for continuous industrial scale production. Bubble based separation systems are preferred over their non-bubble based counterparts primarily because, particle sedimentation velocity is substantially lower than their rise velocity when attached to bubbles. Thus, flotation improves separation by enhancing buoyancy force over sedimentation. Also, unlike sedimentation where some particles cling and remain attached to the flotation cell wall, microbubbles ensure a clean sweep of particles along their path. Another advantage over sedimentation is the sludge moisture content level after harvest. Moisture content measurement (see Fig. 11) of recovered yeast cells showed a reduced water amount with the microflotation-harvested cells than with cells allowed to sediment. As they rise, microbubbles transport the attached flocs and dispatch the floc particles at the liquid–air interface to initiate formation of the cell (sludge) blanket layer. Once formed, microbubbles continue to transport attached flocs to thicken the blanket but also begin to compress and compact this cell blanket layer, thereby thinning it (forming a closed packed bed). When fully thinned, further injection of microbubbles accumulate at the rear of the blanket layer, so that the layer becomes suspended in the foam structure rather than immersed in the liquid continuous phase as is the case with sedimented cells. [9] reported a similar occurrence for algae. Moisture content results for cells harvested with microbubbles is ∼7% less than cells harvested by sedimentation. The significance of this result is obvious when further processing is required. Dewatering is one such example particularly in cases

where yeast cells are needed just as ‘cream yeast’. Alternatively, heating can be employed for cell drying. The difference in moisture content between sedimentation and microflotation-harvested cells can represent significant energy savings especially for large-scale productions. Microflotation facilitates dewatering through thickening and thinning of the sludge blanket. Neutral pH condition yielded lower moisture content than acidic conditions due to the relatively larger and coarser flocs formed under neutral pH. Under alkaline condition, highest cell moisture content was obtained because the gelatinous nature of flocculated cells. For each pH condition however, higher moisture content was measured for cells allowed to separate by sedimentation. Note that only results under pH 5 and 7 for cell recovery with microbubbles are shown here as no significant cell recovery with bubbles was achieved under pH 9. 3.8. Actual fermentation broth Given the series of experiments to be conducted, commercially available yeast cells and culture medium was simulated instead of yeast cells from actual fermentation broth, as the former is readily obtainable. It would require a large volume of actual fermentation broth (∼200 L) to undertake and complete the research study. This was impractical from the logistic point of view but more importantly, is the challenge in maintaining the sterility of pilot scale fermentation broth. To test the suitability of our system on actual fermentation broth however, samples (Algist Bruggeman, Belgium) were obtained and recovery efficiency over three replicates of 95% was obtained. The slight decrease in the separation result may be due to the relatively viscous fermentation broth compared to the simulated YPD medium. This would influence the bubble size adversely and consequently, decrease the bubble hold-up. Provided the appropriate pretreatment process (coagulation and flocculation) microbubble application is a proven process to attach and lift particles from the liquid continuous phase. Microflotation can work with other agglomerating agents. It is true that there are different costs associated with microflotation than centrifugation or sedimentation, but in other sectors, for instance water purification and fine minerals segregation, flotation separations have proved more economical than these methodologies. It is a reasonable expectation that dramatically lower operating and capital costs of microflotation will translate to yeast. 4. Conclusion

Fig. 11. Moisture content results of harvested cells for two recovery techniques: microflotation and sedimentation.

The recovery of yeast from growth medium has been investigated using microflotation. Size, zeta potential and density of flocs are all characteristics that can be affected by both the medium pH

182


and the flocculant concentration. Acidic and neutral conditions are favorable for flotation but alkaline conditions are less so due to the nature of flocs generated. Recovery efficiency is a function of chitosan concentration and varies directly with increased dosage until a critical concentration is attained before decrease in efficiency occurs. Bubble size affect recovery efficiency in a different way. For smaller particles, separation efficiency is more effective with small bubbles because of the increased collision probability. As particle size increases, the tangential stress imposed by the particle on the bubble increases, consequently bringing about a decrease in the terminal rise velocity of the bubble as well as a corresponding increase in its residence time. Therefore, recovery efficiency becomes low. As the bubble size increases, their buoyant force balances out the tangential stress for the given particle size, and as such, optimum recovery efficiency is obtained. Further increase in bubble size results in reduced collision efficiency with particles and as such, low recovery efficiency. Thus, below or beyond the critical bubble–particle ratio, efficiency becomes less optimal. Acknowledgements WZ would like to acknowledge support from the Concept Fund of Yorkshire Forward and the EPSRC (grant no. EP/I019790/1). WZ would like to acknowledge the Royal Society for a Brian Mercer Innovation award and the Royal Academy of Engineering for an industrial secondment with AECOM Design Build. JOH would like to thank the University of Sheffield for a doctoral scholarship. Helpful discussions and support from Dexu Kong and Peter Sandfort of KATZEN Int’l are acknowledged. Many thanks also to Vaclav Tesar for helpful discussions. CB and E wish to acknowledge the UK Engineering and Physical Sciences Research Council (EPSRC) for a studentship for Karunakaran, Advanced Research Fellowship for Biggs (EP/E053556/01) and further project funding (EP/E036252/1). References [1] V. Boussinesq, Sur la resistance qu’oppose un liquide indefini en repos, in: Acad Sci, 1885, pp. 935–937. [2] W.P. Cheng, F.H. Chi, R.F. Yu, Y.C. Lee, Using chitosan as a coagulant in recovery of organic matters from the mash and lauter wastewater of brewery, Journal of Polymers and the Environment 13 (2005) 383–388. [3] Z. Dai, D. Fornasiero, J. Ralston, Particle–bubble collision models – a review, Advances in Colloid and Interface Science 85 (2000) 231–256. [4] A.L. Demain, H.J. Phaff, C.P. Kurtzman, The industrial and agricultural significance of yeasts, in: P.K. Cletus, W.F. Jack (Eds.), The Yeasts, Fourth Ed., Elsevier, Amsterdam, 1998, pp. 13–19 (Chapter 3). [5] R. Divakaran, V. Sivasankara Pillai, Flocculation of kaolinite suspensions in water by chitosan, Water Research 35 (2001) 3904–3908.

[6] J.K. Edzwald, Dissolved air flotation and me, Water Research 44 (2010) 2077–2106. [7] G. Fleet, in: A. Querol, G. Fleet (Eds.), The Commercial and Community Significance of Yeasts in Food and Beverage Production: Yeasts in Food and Beverages, Springer, Berlin/Heidelberg, 2006, pp. 1–12. [8] R.J. Gochin, J. Solari, The role of hydrophobicity in dissolved air flotation, Water Research 17 (1983) 651–657. [9] J. Hanotu, H.C.H. Bandulasena, W.B. Zimmerman, Microflotation performance for algal separation, Biotechnology and Bioengineering 109 (2012) 1663–1673. [10] J. Hanotu, H.C.H. Bandulasena, T.Y. Chiu, W.B. Zimmerman, Microflotation for oil emulsion separation, International Journal of Multiphase Flow 56 (2013) 119–125. [11] J. Hanotu, K. Ying, O.I. Shada, H. Bandulasena, W.B. Zimmerman, Microalgae recovery by microflotation for biofuel production using metallic coagulants, Biofuels 4 (2013) 363–369. [12] R. Hatti-Kaul, Downstream processing in industrial biotechnology, in: Industrial Biotechnology, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2010, pp. 279–321. [13] D. Hewitt, D. Fornasiero, J. Ralston, Bubble particle attachment efficiency, Minerals Engineering 7 (1994) 657–665. [14] Z. Huang, D. Legendre, P. Guiraud, Effect of interface contamination on particle–bubble collision, Chemical Engineering Science 68 (2012) 1–18. [15] M.D. Machado, M.S.F. Santos, C. Gouveia, H.M.V.M. Soares, E.V. Soares, Removal of heavy metals using a brewer’s yeast strain of Saccharomyces cerevisiae: the flocculation as a separation process, Bioresource Technology 99 (2008) 2107–2115. [16] T. Miettinen, J. Ralston, D. Fornasiero, The limits of fine particle flotation, Minerals Engineering 23 (2010) 420–437. [17] E. Molina Grima, E.H. Belarbi, F.G. Acion Fernandez, A. Robles Medina, Y. Chisti, Recovery of microalgal biomass and metabolites: process options and economics, Biotechnology Advances 20 (2003) 491–515. [18] J. Mukherjee, E. Karunakaran, C.A. Biggs, Using a multi-faceted approach to determine the changes in bacterial cell surface properties influenced by a biofilm lifestyle, Biofouling 28 (2012) 1–14. [19] J. Ralston, S.S. Dukhin, The interaction between particles and bubbles, Colloids and Surfaces A: Physicochemical and Engineering Aspects 151 (1999) 3–14. [20] H.H.M. Rijnaarts, W. Norde, J. Lyklema, A.J.B. Zehnder, The isoelectric point of bacteria as an indicator for the presence of cell surface polymers that inhibit adhesion, Colloids and Surfaces B: Biointerfaces 4 (1995) 191–197. [21] P. Strehaiano, F. Ramon-Portugal, P. Taillandier, in: A. Querol, G. Fleet (Eds.), Yeasts as Biocatalysts – Yeasts in Food and Beverages, Springer, Berlin/Heidelberg, 2006, pp. 243–283. [22] M. Takahashi, Zeta potential of microbubbles in aqueous solutions: electrical properties of the gas–water interface, Journal of Physical Chemistry B 109 (2005) 21858–21864. [23] J. Wang, C. Chen, Biosorption of heavy metals by Saccharomyces cerevisiae: a review, Biotechnology Advances 24 (2006) 427–451. [24] R.H. Yoon, The role of hydrodynamic and surface forces in bubble–particle interaction, International Journal of Mineral Processing 58 (2000) 129–143. [25] W.B. Zimmerman, V. Tesaˇr, S. Butler, H.C.H. Bandulasena, Microbubble Generation, Bentham, Oak Park, IL, United States, 2008. [26] W.B. Zimmerman, V. Tesaˇr, H.C.H. Bandulasena, Towards energy efficient nanobubble generation with fluidic oscillation, Current Opinion in Colloid & Interface Science 16 (2011) 350–356. [27] W.B. Zimmerman, M. Zandi, H.C. Hemaka Bandulasena, V. Tesaˇr, D. James Gilmour, K. Ying, Design of an airlift loop bioreactor and pilot scales studies with fluidic oscillator induced microbubbles for growth of a microalgae Dunaliella salina, Applied Energy 88 (2011) 3357–3369. [28] B.V. Derjaguin, S.S. Dukhin, Theory of flotation of small and medium-size particles, Bulletin-Institute of Mining and Metallurgy 651 (1961) 21–246.

Harvesting and dewatering yeast by microflotation

Recommend Documents