- Email: [email protected]
W emulsions using flat metallic membranes and scale-up
Journal of Membrane Science 430 (2013) 140–149
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Geometric parameters influencing production of O/W emulsions using flat metallic membranes and scale-up Miguel A. Sua´rez, Gemma Gutie´rrez, Jose´ Coca, Carmen Pazos n ´n Claverı´a 8, Oviedo 33006, Spain Department of Chemical and Environmental Engineering, University of Oviedo, Julia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 September 2012 Received in revised form 21 November 2012 Accepted 8 December 2012 Available online 20 December 2012
The influence of geometric parameters (e.g., tank dimensions, diameter and position of impeller, continuous phase height-to-tank diameter ratio and impeller-to-tank diameter ratio) on the droplet size of emulsions produced by membrane emulsification has been studied. Two flat metallic membranes with pore sizes of 5 mm and 50 mm were used to prepare oil-in-water (O/W) emulsions in stirred tanks. Results show that the influence of geometric parameters depends on the rotational speed of the impeller. Under optimal geometric conditions, it was possible to obtain emulsions with minimum droplet sizes of 45 mm and 105 mm and span values of 0.60 and 0.53 using membranes with pore sizes of 5 mm and 50 mm, respectively. Scale-up experiments were carried out with different tanks and impellers, while maintaining geometric similarity. Emulsions with the same droplet size distributions were obtained using the impeller tip speed as the scale-up criterion. & 2012 Elsevier B.V. All rights reserved.
Keywords: Membrane emulsification Flat metallic membrane Stirred tank Geometric parameters Scale-up
1. Introduction Emulsions are widely used in the food, pharmaceutical, cosmetic and coating industries. In some specific applications, such as drug delivery, a very narrow droplet size distribution is required to ensure proper activity. Conventional methods for preparing emulsions (e.g., rotor–stator systems, colloid mills) generate polydisperse droplet size distributions. Since its introduction in the 1990s [1–5], membrane emulsification is considered suitable for preparing monodisperse emulsions. One phase (the dispersed phase) is forced by a pressure gradient (transmembrane pressure) to flow through the pores of a membrane into a second phase (the continuous phase). Droplets formed on the membrane surface are detached and dispersed into the continuous phase by applying a low shear stress. Membranes used for emulsification may have different geometries (flat or tubular) and are available in several materials: ceramic [3,5–7], metal [8–11], microporous glass [1,2] and organic polymers [12,13]. Metallic membranes have been used in crossflow [14–16] systems and commercial stirred tank modules [8–11] and also in rotating and oscillating devices [17–19]. The main advantage of metallic membranes is their mechanical strength, compared to the more fragile ceramic and microporous membranes. In addition metallic membranes produce a narrower droplet size distribution. Control of the droplet size coupled with an associated narrow size distribution is a critical factor in membrane emulsification. Several parameters (see Table 1), influence both factors: emulsion composition, membrane type, equipment characteristics and
n
Corresponding author. Tel.: þ34 985103509; fax: þ 34 985103434. E-mail address: [email protected] (C. Pazos).
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.12.013
operating parameters [20–22]. The influence of membrane type [3,23,24] and operating parameters [20,23,25–27] has been studied for several membrane emulsification systems and composition parameters [5,27,28]. However, the composition parameters are determined by the formulation of the emulsion, which often is imposed by the final application of the emulsion, so changes are not always possible. Furthermore, equipment parameters, which are important for scaling-up membrane emulsification processes have not been studied as thoroughly because of the limitations of the laboratory equipment used. Several studies on membrane emulsification in tubular crossflow systems at pilot plant scale have been reported [5,29–32]. In addition, some experimental parallel flow designs have been proposed for microengineered devices in order to increase their throughput [33]. The objective of this work was to study the influence of equipment and operating parameters on droplet size distribution for pilot plant scale membrane emulsification with flat membranes in stirred tanks. An appropriate combination of these parameters might help to make the emulsification system more versatile and to retain a constant droplet size distribution when larger tanks and impellers are used.
2. Experimental 2.1. Materials Emulsions were prepared with a food-grade extra virgin olive oil (md ¼51 mPa s, rd ¼886 kg/m3, and n¼ 1.4677, at 25 1C; acid value lower than 0.8) as the dispersed phase. Emulsions were
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
141
Table 1 Parameters influencing membrane emulsification. Process variables
Degrees of freedom
Properties and design parameters
Composition
Continuous phase (CP) Dispersed phase (DP) Surfactant concentration Concentration of DP Concentration of co-stabilizers
Interfacial tension Viscosities of DP and CP Densities of DP and CP Stability behaviour (coalescence, creaming)
Membrane
Membrane material Membrane shape Pore size and shape
Pore distance Contact angle of DP and CP Porosity Pore tortuosity Membrane area Membrane thickness Emulsification system
Equipment
Emulsification system
Emulsion volume CP pump DP feed system Pipes, circuit Impeller (size and design) Tank dimensions Stirred tank geometric ratios
Operating conditions
CP flow rate Transmembrane pressure DP flux Rotational speed of the impeller
Shear stress Droplet growth time Droplet detachment time
stabilized using a non-ionic surfactant added to the continuous aqueous phase, namely Tween 20s at 2 wt%, along with a viscosity modifier, medium viscosity sodium carboxymethylcellulose with a degree of polymerization 1100 (CMCNa), at 0.01 wt% (both supplied by Sigma Aldrich). 2.2. Methods Emulsification was achieved using a flat membrane module of special design (Fig. 1), which has been described previously [34]. The device has a behaviour similar to that of commercial units operating with small volumes of continuous phase. The module can be used with tanks of different diameters, geometry and material. In addition, it is possible to use impellers of different design and size if other shear rates or flow patterns are desired. Two tailor-made flat membranes fabricated from nickel and stainless steel were used for the experiments. They had a regular array of pores and an active diameter of 3.15 cm. The nickel membrane, supplied by Micropore Ltd. (Derbyshire, United Kingdom), had a 5 mm pore size, with a distance between pores of 200 mm, and a thickness of 200 mm. The stainless steel membrane, supplied by Pantur S.L. (Barcelona, Spain), had a 50 mm pore size, with a distance of 500 mm between pores, and a thickness of 50 mm. Membranes were cleaned with a dishwashing detergent and deionised water in an ultrasound bath for 10 min, followed by rinsing for 10 min with acetone in an ultrasound bath. Finally, they were dried using compressed air and soaked in the continuous phase. The shear stress for droplet detachment was provided by three impellers moved by a Heidolph 2102 RZR motor with rotational speed control. The two paddle impellers were 0.06 m and 0.09 m long and 0.015 m and 0.03 m high, respectively. These impellers were selected because they generate a radial flow pattern [35], suitable for droplet detachment. The third impeller was a marine propeller with three blades, 0.06 m diameter and a pitch of 1.7. The distance between each impeller and the membrane surface was 5 mm. The dispersed phase was gently fed into the cell using a Masterflex peristaltic pump (Cole Parmer Instrumental Co., Chicago). All the membrane pores were active under the operating conditions
and dispersed phase flux was in the range of 17–34 L/m2 h (0.2–0.4 g/min) so that oil concentrations of ca. 1.5 wt% were achieved. All emulsification experiments were conducted at room temperature, and replicated three times. The standard deviation of the droplet size was less than 5 mm, with a coefficient of variation in the range 3–12%. The droplet size distribution was measured using laser diffraction with a Malvern Mastersizer S (Malvern Instruments Ltd, UK). Droplet diameter (Dd) relates to the mean volume diameter (D[4,3]). 2.3. Selection of geometric parameters The droplet size has been considered the control criterion for experiments in which geometric parameters are changed. It is commonly assumed that the smaller the droplet size, the better the performance of the emulsification process. The width of droplet size distribution, given by the span value [(D90 D10)/D50], has also been considered as an additional criterion to control the emulsification process. Paddle impellers have often been used [8–10] because in stirred tanks they provide a radial flow pattern [35] suitable for droplet detachment. In this work, the performance characteristics of two impellers providing radial flow (paddle impellers) and one providing axial flow (a marine propeller) were compared. The influence of impeller diameter was also studied for paddle impellers. Shear stress has been reported to be a key parameter affecting the droplet size of emulsions prepared in stirred tanks. Models for commercial cells [8], which are based on general stirred tank hydrodynamics [36], do not include geometric parameters because they are based on impeller-to-tank ratios [36,37]. The shear stress (t) for Newtonian liquids in the turbulent regime can be determined from the volumetric power input (P/V), Eq. (1). This expression of the shear stress includes geometric parameters [34,38,39]. rffiffiffiffiffiffiffiffiffiffi P t ¼ mc ð1Þ V
mc and V are the continuous phase viscosity and volume, respectively. P/V can be estimated from dimensionless correlations plots [35,40] for different impellers and operating conditions,
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
142
Fig. 1. Schematic diagram of the membrane emulsification cell (A) and the module structure (B).
commonly used in design to estimate power consumption. However, if torque (M) can be measured, an alternative for estimating the shear stress is given in Eq. (2), where DT is the tank diameter, H the height of the continuous phase and N is the rotational speed of the impeller.
t¼
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8mc NM HD2T
ð2Þ
According to Eq. (2) the height of the continuous phase might influence both the droplet size and the hydrostatic pressure on membrane surface as well as the vortex depth. This influence can be expressed in terms of the height-to-tank diameter ratio (H/DT), which is another geometric parameter typically specified in the process design. The impeller position is also important in mixing operations [40], which is given by the distance between the impeller and the bottom of the tank. In this work, the impeller position is expressed in terms of the distance between the impeller and the membrane surface The impeller-to-tank diameter ratio (Di/DT) is a geometric parameter in mixing operations. High ratios (large impellers) provide high fluid flow but a low shear rate (especially because lower rotational speeds are required). These parameters affect the
droplet size. It has also been reported that high Di/DT ratios provide a high torque and low power consumption [40]. In mixing operations there is always an optimum ratio for different impeller design and size. Some geometric ratios or standard dimensions ensure good mixing conditions [35]. However, in membrane emulsification the optimal goal is to attain a monodisperse emulsion with a small droplet size. This goal differs from the optimum for mixing operations (perfect mixing, minimum power consumption). Hence the optimal geometric ratios for preparing emulsions might be different from those for mixing.
3. Results and discussion 3.1. Influence of impeller type A paddle impeller and a marine propeller, both with a diameter of 0.06 m, were used in the experiments. The behaviours of the two impellers were characterized using four different rotational speeds (in the range of 300–600 rpm) with two different membranes: a nickel membrane with a pore size of 5 mm and a stainless steel membrane with a pore size of 50 mm. Droplet sizes obtained with both membranes are shown in Fig. 2.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
3.2. Influence of the height-to-tank diameter ratio of the continuous phase (H/DT)
80
Droplet diameter, Dd (µm)
75 Paddle Marine propeller
70 65 60 55 50 45 40 300
143
350
400 450 500 Rotational speed, N (rpm)
550
600
The influence of this parameter was determined using four different rotational speeds (from 300 rpm to 600 rpm) with a paddle impeller of 0.06 m diameter, and the same membranes (5 mm and 50 mm pore sizes) described in Section 3.1. Four different H/DT ratios were studied: 0.7, 0.85, 1.0 and 1.3, that correspond to continuous phase volumes of 1.0, 1.2, 1.4 and 1.8 L, respectively. In this series of trials the paddle impeller with a diameter of 0.06 m was used in a tank of 0.12 m diameter. Results are shown in Fig. 3. No significant difference in droplet size was observed at low rotational speeds. However, an important effect was observed for both membranes when the rotational speed was increased. At low H/DT values and high rotational speeds, larger droplets are obtained with both membranes. The difference in the mean droplet size was 10 mm for the 5 mm pore membrane and 20 mm for the 50 mm pore membrane. Furthermore, the droplet sizes at H/DT ratios of 1.0 and 1.3 are very similar. Hence, H/DT ratios
180
Droplet diameter, Dd (µm)
160 140 120 100 Paddle Marine propeller
80 60 40 300
350
400 450 500 Rotational speed, N (rpm)
550
600
Fig. 2. Influence of impeller rotational speed on mean droplet size of emulsions prepared with metallic membranes and two types of impellers in the emulsification device, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
The plots in Fig. 2 indicate that at low rotational speeds the paddle impeller generates smaller droplets than the marine propeller. This effect is more pronounced for the membrane with pore size of 5 mm. However, at high rotational speeds there was no substantial difference in mean droplet size. Although paddle impellers provide radial flow, the fluid is also pumped upwards and downwards because of the pressure difference between the two sides of the impeller [40]. This axial flow is similar to that produced by the marine propeller. Hence, droplet sizes do not differ appreciably for the two designs. However, the emulsions produced by the marine propeller were characterized by higher span values (0.69 and 0.61 for the 5 mm and 50 mm pore membranes, respectively) than the paddle impeller (0.60 and 0.55 for the same membranes, respectively). Moreover, the emulsions generated using the marine propeller were more sensitive to rotational speed because the difference between the mean droplet size of emulsions prepared at 300 rpm and 600 rpm was larger than for the paddle impeller. This fact makes the paddle impeller easier to control and thus more convenient to use.
Fig. 3. Influence of impeller rotational speed on mean droplet size of emulsions prepared with metallic membranes at different H/DT values in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.06 m. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
144
larger than 1.0 do not lead to a further decrease in droplet size. Therefore, H/DT ¼1.0 can be considered the optimum value, because it is the lowest volume that can yield the smallest droplet size. This H/DT ratio is one of the standard geometric ratios in mixing operations [35,40]. These results can be explained because the smaller the H/DT ratio, the more likely the vortex falls under the impeller level. The practical implication of this situation is that part of the surface of the impeller is not in contact with the liquid. Hence, the entire surface of the paddle is not used in transferring power to the liquid, the torque is lower, and the shear stress decreases. Moreover, the height of continuous phase above the membrane surface decreases as the rotational speed increases, and hence there is an increase in the transmembrane pressure. However, high H/DT ratios imply an increase in liquid volume and a decrease in shear stress according to Eq. (1), which explains the optimum value at H/DT ¼1.0. Similar behaviour was observed for the span values of the two membranes, as shown in Fig. 4. A steady decrease in span was observed as H/DT increased up to 1, but no significant improvement was obtained by increasing this ratio.
tip speeds (u), defined by Eq. (3). The tip speed is related to the tangential velocity. u ¼ pNDi
ð3Þ
Tip speed is a typical scale-up parameter and values for both impellers are indicated in Table 2 and plotted versus droplet size in Fig. 5. Analysis of the data in Fig. 5 indicates that larger droplets were generated with the larger impeller. Although no significant difference was noted for high tip speeds, the influence of the impeller diameter is important at low tip speeds, especially for the 5 mm pore membrane. Because the impeller height is larger for the Di ¼0.09 m impeller, the bottom of the vortex falls under the level of the large impeller even though the tip speed is equivalent. In this situation power is not transferred to the liquid with an associated increase in droplet size. Span values were almost the same for both impellers employed with the 50 mm pore membrane (0.52 and 0.55, respectively), but
80
3.3. Influence of impeller diameter (Di)
0.70 0.68
5 µm 50 µm
0.66
Droplet diameter, Dd (µm)
The influence of impeller diameter on droplet size was studied using the two metallic membranes with paddle impellers (0.06 m and 0.09 m diameter) in a 0.12 m diameter tank. The droplet size was measured at different rotational speeds (N) for both impellers. However, the speed values were chosen according to their
0.64
70 65 60 55 50 45
0.62 Span
Di = 0.06 m Di = 0.09 m
75
40 0.90
0.60
1.10
0.58 0.56
1.70
1.90
150
0.54
Di = 0.06 m Di = 0.09 m
0.50 0.7
1.1
0.9
1.3
H/DT ratio Fig. 4. Influence of H/DT ratio on span values of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.06 m.
Table 2 Equivalent rotational speeds for the same tip speed for impellers of 0.06 m and 0.09 m diameters. Tip speed (m/s)
Rotational speed (rpm)
Rotational speed (rpm)
Di ¼0.06 m
Di ¼ 0.09 m
300 375 450 600
200 250 300 400
Droplet diameter, Dd (µm)
140
0.52
0.94 1.18 1.42 1.88
1.30 1.50 Tip speed, u (m/s)
130 120 110 100 90 80 0.90
1.10
1.30 1.50 Tip speed, u (m/s)
1.70
1.90
Fig. 5. Influence of tip speed on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with paddle impellers of different diameters, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. H/DT ¼ 1.0. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
higher for the larger impeller (0.73) for the trials with the 5 mm pore membrane relative to the small one (0.60).
3.4. Influence of impeller position (Z) The effect of vertical position of the impeller on droplet size distribution was studied as a function of the distance between the impeller and the membrane surface (Z). The study was carried out with both metallic membranes, an H/DT ratio of 1.0, and a paddle impeller (0.06 m diameter) operating at a tip speed of 0.94 m/s (300 rpm). This speed was chosen because at higher values it was not feasible to study a reasonable number of positions because of the vortex effect. Three different Z values were considered: 5 mm (used in previous experiments), 20 mm and 40 mm. Results are shown in Fig. 6. 160
145
No significant effect of the impeller position was observed on the mean droplet size for either membrane at low speeds. Under the chosen conditions no vortex fall takes place. Furthermore, this parameter did not influence the span values. If higher rotational speeds were used and vortex fall would take place, a similar behaviour to Fig. 3 might be expected. However, it should be noted that using a smaller Z implies a longer distance to top surface of the continuous phase. Hence, a wide range of rotational speeds can be used without the vortex falling below the level of the impeller. Consequently, low values of Z would be more suitable for obtaining improved performance in membrane emulsification.
3.5. Influence of impeller-to-tank diameter ratio (Di/DT) The influence of tank size on droplet size was studied with the same paddle impeller (diameter 0.09 m) in three different tanks. Because one experiment required a large volume of continuous
120
150
100
140
80
130
60 40 5 µm
20
50 µm
0 0
20 40 10 30 Distance to membrane surface, Z (mm)
50
Droplet diameter, Dd (µm)
Droplet diameter, Dd (µm)
140
5 µm 50 µm
120 110 100 90 80 70
Fig. 6. Influence of distance between the membrane surface and the impeller (Z) on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m, H/DT ¼ 1.0, N ¼ 300 rpm.
60 0.3
0.7
0.8
145 Droplet diameter, Dd (µm)
95 90 Droplet diameter, Dd (µm)
0.5 0.6 Di/DT ratio
150
100
85 80 75 70 65
140 u = 0.94 m/s u = 1.42 m/s
135 130 125 120 115 110
60
105
55 50 0.10
0.4
0.15
0.20 0.25 Tank diameter, DT (m)
0.30
Fig. 7. Influence of tank diameter on mean droplet size of emulsions prepared with a 5 mm metallic membrane with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.09 m, H/DT ¼ 1.0, N ¼ 200 rpm.
100 0.3
0.4
0.5 0.6 Di/DT ratio
0.7
0.8
Fig. 8. Influence of impeller-to-tank diameter ratio on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. H/DT ¼ 1.0. (A) u¼ 0.94 m/s, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
146
phase, the less favourable conditions reported in Section 3.3 were employed to minimize surfactant consumption. These trials utilized the 5 mm pore membrane, and a tip speed of 0.94 m/s (200 rpm). The three tanks had diameters of 0.12 m, 0.18 m and 0.30 m. These diameters correspond to volumes of 1.4 L, 4.5 L and 25 L, respectively. All were employed at the optimum H/DT ratio (1.0) and results are shown in Fig. 7. The difference in droplet size between the smallest and the largest tank is less than 5 mm. Consequently, an increase in tank size does not lead to worse performance. The smallest droplet size corresponds to emulsions prepared in a 0.18 m diameter tank. For this impeller/tank combination Di/DT ¼0.5. No influence was observed in the span values (0.65 70.05). The experiments represented in Fig. 7 correspond to three different impeller-to-tank diameter ratios, i.e., 0.33, 0.5 and 0.75. The experiments in Fig. 8 correspond to both membranes, using the same impeller-to-tank diameter ratios and a tip speed of 0.94 m/s (see Fig. 8A). In addition, another tip speed (1.42 m/s) was employed in trials with the membrane of 50 mm pore size (see Fig. 8B). The mean droplet size of emulsions obtained at a tip speed of 0.94 m/s varied by only a few percent. The maximum value was obtained at Di/DT ratio of 0.75. However, different behaviour was observed for the trials at a tip speed of 1.42 m/s (Fig. 8B). The mean droplet size was significantly smaller for low Di/DT ratios. This result might be explained by the vortex fall at high impeller rotational speed, so that part of the impeller does not transfer power to the liquid. This vortex fall did not take place at a tip speed of 0.94 m/s. In addition, it has been reported [40] that the higher the Di/DT ratio the lower the average shear rate, so that the droplet size should increase. The standard ratio in mixing operations is Di/DT ¼ 0.3; however, the optimum ratio for membrane emulsification depends on the type of impeller and operating conditions. The span values were relatively invariant at values between 0.50 and 0.55.
3.6. Influence of baffles Vortex formation was observed in experiments at high rotational speeds. Vortexing was a problem, because power was not completely transferred to the continuous phase. It has been reported [35,41] that the vortex is caused by the tangential component of the continuous phase velocity. This factor is also related to droplet detachment [3]. To determine the influence of tangential velocity and vortex, experiments were performed in a tank with and without baffles using both metallic membranes (5 mm and 50 mm pores) with the 0.06 m diameter impeller at a rotational speed of 600 rpm. The baffles had a height of 0.17 m and a width of 0.014 m. Results are shown in Fig. 9, for several emulsification times, t. It is observed in Fig. 9A that the differences in the droplet size distribution are not very significant for the 5 mm pore size membrane and that the droplet size is slightly larger in the baffled tank. However, the droplet size distribution obtained after 30 min with the 50 mm pore size membrane was significantly different. The droplets were smaller and the size distribution was wider in the baffled tank. The evolution of the emulsion with time was studied in the baffled tank. Breakup of droplets occurred for the 50 mm membrane, as shown in Fig. 9B, but it was not noticed for the 5 mm membrane (Fig. 9C). Moreover, no breakup was observed in the unbaffled tank (Fig. 9D). This fact could be explained because of the large size of the droplets obtained with the 50 mm membrane. Such droplets are more sensitive to shear in the vicinity of the baffles [35], yielding a wider droplet size distribution. 3.7. Membrane emulsification in large-volume tanks The purpose of scale-up of membrane emulsification is to obtain emulsions at large scale having the same droplet size, but
30
30
20 15
25
5 µm unbaffled 5 µm baffled 50 µm unbaffled 50 µm baffled
Volume (%)
Volume (%)
25
10 5 0 0.01
0.1
1 10 Diameter (µm)
100
15
0.1
1 10 Diameter (µm)
100
1000
100
1000
30 Baffled t = 10 min Baffled t = 20 min Baffled t = 30 min Unbaffled t = 30 min
25
10 5 0 0.01
10
0 0.01
1000
Volume (%)
Volume (%)
20
15
5
30 25
20
Baffled t = 10 min Baffled t = 20 min Baffled t = 30 min Unbaffled t = 30 min
t = 30 min t = 45 min t = 60 min
20 15 10 5
0.1
1 10 Diameter (µm)
100
1000
0 0.01
0.1
1 10 Diameter (µm)
Fig. 9. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m, DT ¼0.15 m, H/DT ¼1.0, N ¼ 600 rpm. (A) t¼ 30 min, (B) 50 mm pore size membrane, (C) 5 mm pore size membrane (D) 50 mm pore size membrane and unbaffled tank.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
Table 3 Scale-up criteria for stirred tanks. Scale-up parameter
Equation
Tip speed
u ¼ pNDi
Reynolds number
Re ¼ P ¼ V
Power per unit volume
Conditions in the larger tank ð4Þ
rc NDi 2 ð5Þ mc 2pNM V
Di1 Di2 Di1 2 N2 ¼ N1 Di2 3 Di2 P2 ¼ P 1 Di1 N2 ¼ N1
ð6Þ
Table 4 Conditions for scaling-up membrane emulsification using paddle impellers (0.06 and 0.09 m diameter) in stirred tanks (1.4 L and 4.5 L continuous phase) keeping Di/DT ¼ 0.5. Parameter
Small tank (1.4 L continuous phase)
Large tank (4.5 L continuous phase)
Tip speed Reynolds number Power per unit volume
N1 ¼ 300 rpm N1 ¼ 300 rpm P1 ¼ 1.08 W (N1 ¼ 300 rpm)
N2 ¼ 200 rpm N2 ¼ 133 rpm P2 ¼3.65 W (N2 4400 rpm)
ð7Þ ð8Þ
147
ð9Þ
35 30 25
V = 1.4L; u = 1.42 m/s V = 4.5L; u = 1.42 m/s
30
V = 4.5L V = 25L
25
20
Volume (%)
Volume (%)
35
V = 1.4L; u = 0.94 m/s V = 4.5L; u = 0.94 m/s
15
20 15
10 10
5 5
0 0.1
1
10 Diameter (µm)
100
1000
0 0.1
1
10 Diameter (µm)
100
1000
40 35
Fig. 11. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di/DT ¼ 0.33, H/DT ¼ 1.0, u¼ 0.94 m/s.
V = 1.4L; u = 0.94 m/s V = 4.5L; u = 0.94 m/s
Volume (%)
30 25
V = 1.4L; u = 1.42 m/s V = 4.5L; u = 1.42 m/s
20 15 10 5 0 0.1
1
10 Diameter (µm)
100
1000
Fig. 10. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di/DT ¼ 0.5, H/DT ¼ 1.0. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
maintaining at least geometric similarity. Different criteria for scaling-up involving stirred tank mixing have been considered and the equations (Eq. (4)–(9)) are summarized in Table 3, in which the subscript ‘‘1’’ refers to the laboratory (small) unit and ‘‘2’’ to the larger scale unit. Experiments were conducted in two stirred tanks with diameters of 0.12 m and 0.18 m. These diameters correspond to continuous phase volumes of 1.4 L and 4.5 L, respectively. Two paddle impellers with diameters of 0.06 m and 0.09 m were
utilized, H/DT and Di/DT ratios were set to 1.0 and 0.5, respectively, and two impeller tip speeds were used (0.94 m/s and 1.42 m/s) along with flat metallic membranes with pore sizes of 5 mm and 50 mm. The scale-up parameters in Table 3 were explored in the experimental trials. Tip speed was the first criterion used and results are shown in Fig. 10. The droplet size distributions of emulsions prepared at the same tip speed in different tanks, maintaining geometric similarity, are the same. Data were obtained at two different tip speeds and for the two membranes with different pore sizes. The conditions for using P/V and Reynolds number as scale-up criteria are summarized in Table 4. The volumetric power input, P/V, commonly used to scale-up mixing in stirred tanks, does not seem to be a suitable criterion for scale-up of membrane emulsification because it implies using very high rotational speeds. The entries in Table 4 for scaling-up experiments at 300 rpm in the small tank would need using rotational speeds higher than 400 rpm in the large tank. However, even at a rotational speed of 300 rpm, a different droplet size is produced. Neither is the Reynolds number a good criterion for scale-up, because the rotational speed for the large tank is underestimated. In order to verify the suitability of tip speed as scale-up criterion, which is considered as non-conservative for mixing operations, another experiment was carried out in a larger tank with 25 L of continuous phase (Di ¼ 0.09 m, Di/DT ¼0.33 and H/DT ¼ 1.0). The results were compared with those obtained in the tank with 4.5 L of continuous phase. The 5 mm pore size
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
148
membrane was used and rotational speed was set to 300 rpm in the smaller tank. Droplet size distributions are shown in Fig. 11. If geometric ratios and tip speed are kept constant a significant increase in tank size does not affect droplet size. Even though tip speed is not considered the best scale-up parameter in mixing operations, it seems to be appropriate for membrane emulsification. This criterion could also be expressed in terms of dimensionless numbers containing tip speed, such as the capillary and Euler numbers [34]. However, diffusion, which has been reported to affect membrane emulsification [42], does not seem to have much influence under the experimental conditions and for the emulsion used in this work. If the process were affected by diffusion, other scale-up parameters related to mass transfer (e.g., the mass transfer coefficient) [41] would be more suitable. It has been also reported that tip speed is related to the maximum shear rate [40], which could mean that the maximum shear stress is the key parameter in determining droplet size during membrane emulsification. It should also be noted that the membrane active area in the emulsification unit shown in Fig. 1 is placed within the forced vortex region, where shear stress is not affected by the tank walls [8,35,41]. Thus, droplet detachment might be different if the membrane active area were increased.
D90 D[4,3] Dd Di Di/DT DT H H/DT M n N P P/V Re t u V Z
Diameter for which 90% of distribution volume has smaller size (mm) Diameter of the sphere with volume equivalent to that of the droplet (mm) Mean droplet diameter (mm) Impeller diameter (m) Impeller-to-tank diameter ratio Tank diameter (m) Height of liquid (continuous phase) in the tank (m) Continuous phase height-to-tank diameter ratio Torque (N m) Refractive index Rotational speed (rps) Power input (W) Power per unit volume (W/m3) Reynolds number for stirring time (s) Impeller tip speed (m/s) Volume of liquid (continuous phase) in the tank (m3) Distance from impeller to membrane surface (m)
Greek letters 4. Conclusions Membrane emulsification experiments to produce O/W emulsions, reported in this work, lead to the following conclusions:
Paddle impellers exhibit better performance than marine
propellers for membrane emulsification because the resulting emulsions are less sensitive to rotational speed. However, differences are higher at low rotational speeds. Low H/DT ratios lead to large droplets, especially for high rotational speeds. The optimum value was H/DT ¼1.0, that corresponds to standard ratios in mixing operations. Furthermore, the Di/DT ratio has little influence on droplet size, nor do the diameters of the impeller and the tank. The distance between the impeller and the membrane surface does not affect the droplet size of the emulsion at low speeds. However, when baffles are used droplets are larger than in an unbaffled tank. In addition, breakup of the droplets was observed for large droplets in baffled tanks. Impeller tip speed is a key parameter for scaling-up, because at a certain value the same droplet size distributions were obtained, when geometric similarity was maintained. Finally, optimum conditions in mixing operations are not necessarily the optimum conditions for membrane emulsification.
Acknowledgements This work was supported by the MICINN, Spain, under the grants CTQ2007-65348 and MICINN-10-CTQ2010-20009-C02-01. M.A. Sua´rez acknowledges receipt of a graduate fellowship from FPU Program (MECD, Spain), co-funded by European Social Fund. The authors thank Prof. Charles Hill for fruitful discussions.
Nomenclature D10 D50
Diameter for which 10% of distribution volume has smaller size (mm) Diameter for which 50% of distribution volume has smaller size (mm)
g mc md rc rd t
Shear rate (s 1) Continuous phase dynamic viscosity (Pa s) Dispersed phase dynamic viscosity (Pa s) Continuous phase density (kg/m3) Dispersed phase density (kg/m3) Shear stress (Pa)
Subscripts 1 2
Small tank Large tank
References [1] T. Nakashima, M. Shimizu, M. Kukizaki, Membrane emulsification by microporous glass, Key Eng. Mater. 61&62 (1991) 513–516. [2] T. Nakashima, M. Shimizu, M. Kukizaki, Particle control of emulsion by membrane emulsification and its applications, Adv. Drug Delivery Rev. 45 (2000) 47–56. [3] S.J. Peng, R.A. Williams, Controlled production of emulsions using a crossflow membrane: Part I: Droplet formation from a single pore, Chem. Eng. Res. Des. 76 (1998) 894–901. ¨ [4] V. Schroder, O. Behrend, H. Schubert, Effect of dynamic interfacial tension on the emulsification process using microporous ceramic membranes, J. Colloid Interface Sci. 202 (1998) 334–340. ¨ [5] V. Schroder, H. Schubert, Production of emulsions using microporous ceramic membranes, Colloids Surf., A: Physicochem. Eng. Aspects 152 (1999) 103–109. [6] C. Charcosset, H. Fessi, Preparation of nanoparticles with a membrane contactor, J. Membr. Sci. 266 (2006) 115–120. [7] N. Sheibat-Othman, T. Burne, C. Charcosset, H. Fessi, Preparation of pHsensitive particles by membrane contactor, Colloids Surf., A: Physicochem. Eng. Aspects 315 (2008) 13–22. [8] S.R. Kosvintsev, G. Gasparini, R.G. Holdich, I.W. Cumming, M.T. Stillwell, Liquid–liquid membrane dispersion in a stirred cell with and without controlled shear, Ind. Eng. Chem. Res. 44 (2005) 9323–9330. [9] M.T. Stillwell, R.G. Holdich, S.R. Kosvintsev, G. Gasparini, I.W. Cumming, Stirred cell membrane emulsification and factors influencing dispersion drop size and uniformity, Ind. Eng. Chem. Res. 46 (2007) 965–972. [10] M.M. Dragosavac, M.N. Sovilj, S.R. Kosvintsev, R.G. Holdich, G.T. Vladisavljevic, Controlled production of oil-in-water emulsions containing unrefined pumpkin seed oil using stirred cell membrane emulsification, J. Membr. Sci. 322 (2008) 178–188. [11] E. Egidi, G. Gasparini, R.G. Holdich, G.T. Vladisavljevic, S. Kosvintsev, Membrane emulsification using membranes of regular pore spacing: Droplet size
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22] [23]
[24]
[25] [26]
and uniformity in the presence of surface shear, J. Membr. Sci. 323 (2008) 414–420. I. Kobayashi, M. Yasuno, S. Iwamoto, A. Shono, K. Satoh, M. Nakajima, Microscopic observation of emulsion droplet formation from a polycarbonate membrane, Colloids Surf., A: Physicochem. Eng. Aspects 207 (2002) 185–196. ¨ A. Trentin, C. Guell, F. Lo´pez, M. Ferrando, Microfiltration membranes to produce BSA-stabilized emulsions by premix membrane emulsification, J. Membr. Sci. 356 (2010) 22–32. P.J. Dowding, J.W. Goodwin, B. Vincent, Production of porous suspension polymer beads with a narrow size distribution using a cross-flow membrane and a continuous tubular reactor, Colloids Surf., A: Physicochem. Eng. Aspects 180 (2001) 301–309. M.J. Geerken, R.G.H. Lammertink, M. Wessling, Interfacial aspects of water drop formation at micro-engineered orifices, J. Colloid Interface Sci. 312 (2007) 460–469. M.J. Geerken, M.N.W. Groenendijk, R.G.H. Lammertink, M. Wessling, Microfabricated metal nozzle plates used for water-in-oil and oil-in-water emulsification, J. Membr. Sci. 310 (2008) 374–383. V. Schadler, E.J. Windhab, Continuous membrane emulsification by using a membrane system with controlled pore distance, Desalination 189 (2006) 130–135. G.T. Vladisavljevic, R.A. Williams, Manufacture of large uniform droplets using rotating membrane emulsification, J. Colloid Interface Sci. 299 (2006) 396–402. R.G. Holdich, M.M. Dragosavac, G.T. Vladisavljevic, S.T. Kosvintsev, Membrane emulsification with oscillating and stationary membranes, Ind. Eng. Chem. Res. 49 (2010) 3810–3817. A.J. Gijsbertsen-Abrahamse, A. van der Padt, R.M. Boom, Status of cross-flow membrane emulsification and outlook for industrial application, J. Membr. Sci. 230 (2004) 149–159. D.X. Hao, F.L. Gong, G.H. Hu, Y.J. Zhao, G.P. Lian, G.H. Ma, Z. Su, Controlling factors on droplets uniformity in membrane emulsification: experiment and modeling analysis, Ind. Eng. Chem. Res. 47 (2008) 6418–6425. M.A. Suarez, J. Coca, C. Pazos, Membrane emulsification: factors influencing the size and distribution of drops, Ing. Quim. 505 (2012) 58–73. G.T. Vladisavljevic, H. Schubert, Influence of process parameters on droplet size distribution in SPG membrane emulsification and stability of prepared emulsion droplets, J. Membr. Sci. 225 (2003) 15–23. G.T. Vladisavljevic, U. Lambrich, M. Nakajima, H. Schubert, Production of O/W emulsions using SPG membranes, ceramic–aluminium oxide membranes, microfluidizer and a silicon microchannel plate-a comparative study, Colloids Surf., A: Physicochem. Eng. Aspects 232 (2004) 199–207. ¨ ard, ˚ S.M. Joscelyne, G. Trag Membrane emulsification—a literature review, J. Membr. Sci. 169 (2000) 107–117. G.T. Vladisavljevic, R.A. Williams, Recent developments in manufacturing emulsions and particulate products using membranes, Adv. Colloid Interface Sci. 113 (2005) 1–20.
149
[27] M. Kukizaki, Shirasu porous glass (SPG) membrane emulsification in the absence of shear flow at the membrane surface: influence of surfactant type and concentration, viscosities of dispersed and continuous phases, and transmembrane pressure, J. Membr. Sci. 327 (2009) 234–243. [28] G. Vladisavljevic, M. Shimizu, T. Nakashima, Production of multiple emulsions for drug delivery systems by repeated SPG membrane homogenization: Influence of mean pore size, interfacial tension and continuous phase viscosity, J. Membr. Sci. 284 (2006) 373–383. [29] R.A. Williams, S.J. Peng, D.A. Wheeler, N.C. Morley, D. Taylor, M. Whalley, D.W. Houldsworth, Controlled production of emulsions using a crossflow membrane. Part II: Industrial scale manufacture, Chem. Eng. Res. Des. 76 (1998) 902–910. [30] I. Limayem Blouza, C. Charcosset, S. Sfar, H. Fessi, Preparation and characterization of spironolactone-loaded nanocapsules for paediatric use, Int. J. Pharm. 325 (2006) 124–131. [31] Q. Yuan, R.A. Williams, Large scale manufacture of magnetic polymer particles using membranes and microfluidic devices, China Part 5 (2007) 26–42. [32] E. Lepercq-Bost, M.L. Giorgi, A. Isambert, C. Arnaud, Use of the capillary number for the prediction of droplet size in membrane emulsification, J. Membr. Sci. 314 (2008) 76–89. ¨ R.M. Boom, Parallelized edge-based [33] K.C. v. Dijke, G. Veldhuis, K. Schroen, droplet generation devices, Lab Chip 9 (2009) 2824–2830. [34] M.A. Suarez, G. Gutierrez, J. Coca, C. Pazos, Stirred tank membrane emulsification using flat metallic membranes: a dimensional analysis, Chem. Eng. Process, Unpublished results. [35] F.A. Holland, F.S. Chapman, Liquid Mixing and Processing in Stirred Tanks, Reinhold Publishing, New York, 1966. [36] S. Nagata, Mixing. Principles and Applications, John Wiley & Sons Inc., New York, 1975. [37] M. Ciofalo, Turbulent flow in closed and free surface unbaffled tanks stirred by radial impellers, Chem. Eng. Sci. 51 (14) (1996) 3557–3573. [38] H.J. Henzler, J. Kauling, Scale up of mass transfer in highly viscous liquids, in: ¨ European Conference on Mixing. Wurzburg, Germany, Cranfield BHRA, 1985, pp. 303–312. [39] J.A. Sa´nchez Pe´rez, E.M. Rodrı´guez Porcel, J.L. Casas Lo´pez, J.M. Ferna´ndez Sevilla, Y. Chisti, Shear rate in stirred tank and bubble column bioreactors, Chem. Eng. J. 124 (2006) 1–5. [40] J.Y. Oldshue, Fluid Mixing Technology, McGraw Hill Publications Co, New York, 1983. [41] Z. Sterbacek, P. Tausk, Mixing in the Chemical Industry, Pergamon Press, Oxford, 1965. ¨ ard, ˚ ¨ ard, ˚ [42] M. Rayner, G. Trag C. Trag The impact of mass transfer and interfacial tension expansion rate on droplet size in membrane emulsification processes, Colloids Surf., A: Physicochem. Eng. Aspects 266 (2005) 1–17.
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Geometric parameters influencing production of O/W emulsions using flat metallic membranes and scale-up Miguel A. Sua´rez, Gemma Gutie´rrez, Jose´ Coca, Carmen Pazos n ´n Claverı´a 8, Oviedo 33006, Spain Department of Chemical and Environmental Engineering, University of Oviedo, Julia
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 September 2012 Received in revised form 21 November 2012 Accepted 8 December 2012 Available online 20 December 2012
The influence of geometric parameters (e.g., tank dimensions, diameter and position of impeller, continuous phase height-to-tank diameter ratio and impeller-to-tank diameter ratio) on the droplet size of emulsions produced by membrane emulsification has been studied. Two flat metallic membranes with pore sizes of 5 mm and 50 mm were used to prepare oil-in-water (O/W) emulsions in stirred tanks. Results show that the influence of geometric parameters depends on the rotational speed of the impeller. Under optimal geometric conditions, it was possible to obtain emulsions with minimum droplet sizes of 45 mm and 105 mm and span values of 0.60 and 0.53 using membranes with pore sizes of 5 mm and 50 mm, respectively. Scale-up experiments were carried out with different tanks and impellers, while maintaining geometric similarity. Emulsions with the same droplet size distributions were obtained using the impeller tip speed as the scale-up criterion. & 2012 Elsevier B.V. All rights reserved.
Keywords: Membrane emulsification Flat metallic membrane Stirred tank Geometric parameters Scale-up
1. Introduction Emulsions are widely used in the food, pharmaceutical, cosmetic and coating industries. In some specific applications, such as drug delivery, a very narrow droplet size distribution is required to ensure proper activity. Conventional methods for preparing emulsions (e.g., rotor–stator systems, colloid mills) generate polydisperse droplet size distributions. Since its introduction in the 1990s [1–5], membrane emulsification is considered suitable for preparing monodisperse emulsions. One phase (the dispersed phase) is forced by a pressure gradient (transmembrane pressure) to flow through the pores of a membrane into a second phase (the continuous phase). Droplets formed on the membrane surface are detached and dispersed into the continuous phase by applying a low shear stress. Membranes used for emulsification may have different geometries (flat or tubular) and are available in several materials: ceramic [3,5–7], metal [8–11], microporous glass [1,2] and organic polymers [12,13]. Metallic membranes have been used in crossflow [14–16] systems and commercial stirred tank modules [8–11] and also in rotating and oscillating devices [17–19]. The main advantage of metallic membranes is their mechanical strength, compared to the more fragile ceramic and microporous membranes. In addition metallic membranes produce a narrower droplet size distribution. Control of the droplet size coupled with an associated narrow size distribution is a critical factor in membrane emulsification. Several parameters (see Table 1), influence both factors: emulsion composition, membrane type, equipment characteristics and
n
Corresponding author. Tel.: þ34 985103509; fax: þ 34 985103434. E-mail address: [email protected] (C. Pazos).
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.12.013
operating parameters [20–22]. The influence of membrane type [3,23,24] and operating parameters [20,23,25–27] has been studied for several membrane emulsification systems and composition parameters [5,27,28]. However, the composition parameters are determined by the formulation of the emulsion, which often is imposed by the final application of the emulsion, so changes are not always possible. Furthermore, equipment parameters, which are important for scaling-up membrane emulsification processes have not been studied as thoroughly because of the limitations of the laboratory equipment used. Several studies on membrane emulsification in tubular crossflow systems at pilot plant scale have been reported [5,29–32]. In addition, some experimental parallel flow designs have been proposed for microengineered devices in order to increase their throughput [33]. The objective of this work was to study the influence of equipment and operating parameters on droplet size distribution for pilot plant scale membrane emulsification with flat membranes in stirred tanks. An appropriate combination of these parameters might help to make the emulsification system more versatile and to retain a constant droplet size distribution when larger tanks and impellers are used.
2. Experimental 2.1. Materials Emulsions were prepared with a food-grade extra virgin olive oil (md ¼51 mPa s, rd ¼886 kg/m3, and n¼ 1.4677, at 25 1C; acid value lower than 0.8) as the dispersed phase. Emulsions were
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
141
Table 1 Parameters influencing membrane emulsification. Process variables
Degrees of freedom
Properties and design parameters
Composition
Continuous phase (CP) Dispersed phase (DP) Surfactant concentration Concentration of DP Concentration of co-stabilizers
Interfacial tension Viscosities of DP and CP Densities of DP and CP Stability behaviour (coalescence, creaming)
Membrane
Membrane material Membrane shape Pore size and shape
Pore distance Contact angle of DP and CP Porosity Pore tortuosity Membrane area Membrane thickness Emulsification system
Equipment
Emulsification system
Emulsion volume CP pump DP feed system Pipes, circuit Impeller (size and design) Tank dimensions Stirred tank geometric ratios
Operating conditions
CP flow rate Transmembrane pressure DP flux Rotational speed of the impeller
Shear stress Droplet growth time Droplet detachment time
stabilized using a non-ionic surfactant added to the continuous aqueous phase, namely Tween 20s at 2 wt%, along with a viscosity modifier, medium viscosity sodium carboxymethylcellulose with a degree of polymerization 1100 (CMCNa), at 0.01 wt% (both supplied by Sigma Aldrich). 2.2. Methods Emulsification was achieved using a flat membrane module of special design (Fig. 1), which has been described previously [34]. The device has a behaviour similar to that of commercial units operating with small volumes of continuous phase. The module can be used with tanks of different diameters, geometry and material. In addition, it is possible to use impellers of different design and size if other shear rates or flow patterns are desired. Two tailor-made flat membranes fabricated from nickel and stainless steel were used for the experiments. They had a regular array of pores and an active diameter of 3.15 cm. The nickel membrane, supplied by Micropore Ltd. (Derbyshire, United Kingdom), had a 5 mm pore size, with a distance between pores of 200 mm, and a thickness of 200 mm. The stainless steel membrane, supplied by Pantur S.L. (Barcelona, Spain), had a 50 mm pore size, with a distance of 500 mm between pores, and a thickness of 50 mm. Membranes were cleaned with a dishwashing detergent and deionised water in an ultrasound bath for 10 min, followed by rinsing for 10 min with acetone in an ultrasound bath. Finally, they were dried using compressed air and soaked in the continuous phase. The shear stress for droplet detachment was provided by three impellers moved by a Heidolph 2102 RZR motor with rotational speed control. The two paddle impellers were 0.06 m and 0.09 m long and 0.015 m and 0.03 m high, respectively. These impellers were selected because they generate a radial flow pattern [35], suitable for droplet detachment. The third impeller was a marine propeller with three blades, 0.06 m diameter and a pitch of 1.7. The distance between each impeller and the membrane surface was 5 mm. The dispersed phase was gently fed into the cell using a Masterflex peristaltic pump (Cole Parmer Instrumental Co., Chicago). All the membrane pores were active under the operating conditions
and dispersed phase flux was in the range of 17–34 L/m2 h (0.2–0.4 g/min) so that oil concentrations of ca. 1.5 wt% were achieved. All emulsification experiments were conducted at room temperature, and replicated three times. The standard deviation of the droplet size was less than 5 mm, with a coefficient of variation in the range 3–12%. The droplet size distribution was measured using laser diffraction with a Malvern Mastersizer S (Malvern Instruments Ltd, UK). Droplet diameter (Dd) relates to the mean volume diameter (D[4,3]). 2.3. Selection of geometric parameters The droplet size has been considered the control criterion for experiments in which geometric parameters are changed. It is commonly assumed that the smaller the droplet size, the better the performance of the emulsification process. The width of droplet size distribution, given by the span value [(D90 D10)/D50], has also been considered as an additional criterion to control the emulsification process. Paddle impellers have often been used [8–10] because in stirred tanks they provide a radial flow pattern [35] suitable for droplet detachment. In this work, the performance characteristics of two impellers providing radial flow (paddle impellers) and one providing axial flow (a marine propeller) were compared. The influence of impeller diameter was also studied for paddle impellers. Shear stress has been reported to be a key parameter affecting the droplet size of emulsions prepared in stirred tanks. Models for commercial cells [8], which are based on general stirred tank hydrodynamics [36], do not include geometric parameters because they are based on impeller-to-tank ratios [36,37]. The shear stress (t) for Newtonian liquids in the turbulent regime can be determined from the volumetric power input (P/V), Eq. (1). This expression of the shear stress includes geometric parameters [34,38,39]. rffiffiffiffiffiffiffiffiffiffi P t ¼ mc ð1Þ V
mc and V are the continuous phase viscosity and volume, respectively. P/V can be estimated from dimensionless correlations plots [35,40] for different impellers and operating conditions,
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
142
Fig. 1. Schematic diagram of the membrane emulsification cell (A) and the module structure (B).
commonly used in design to estimate power consumption. However, if torque (M) can be measured, an alternative for estimating the shear stress is given in Eq. (2), where DT is the tank diameter, H the height of the continuous phase and N is the rotational speed of the impeller.
t¼
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8mc NM HD2T
ð2Þ
According to Eq. (2) the height of the continuous phase might influence both the droplet size and the hydrostatic pressure on membrane surface as well as the vortex depth. This influence can be expressed in terms of the height-to-tank diameter ratio (H/DT), which is another geometric parameter typically specified in the process design. The impeller position is also important in mixing operations [40], which is given by the distance between the impeller and the bottom of the tank. In this work, the impeller position is expressed in terms of the distance between the impeller and the membrane surface The impeller-to-tank diameter ratio (Di/DT) is a geometric parameter in mixing operations. High ratios (large impellers) provide high fluid flow but a low shear rate (especially because lower rotational speeds are required). These parameters affect the
droplet size. It has also been reported that high Di/DT ratios provide a high torque and low power consumption [40]. In mixing operations there is always an optimum ratio for different impeller design and size. Some geometric ratios or standard dimensions ensure good mixing conditions [35]. However, in membrane emulsification the optimal goal is to attain a monodisperse emulsion with a small droplet size. This goal differs from the optimum for mixing operations (perfect mixing, minimum power consumption). Hence the optimal geometric ratios for preparing emulsions might be different from those for mixing.
3. Results and discussion 3.1. Influence of impeller type A paddle impeller and a marine propeller, both with a diameter of 0.06 m, were used in the experiments. The behaviours of the two impellers were characterized using four different rotational speeds (in the range of 300–600 rpm) with two different membranes: a nickel membrane with a pore size of 5 mm and a stainless steel membrane with a pore size of 50 mm. Droplet sizes obtained with both membranes are shown in Fig. 2.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
3.2. Influence of the height-to-tank diameter ratio of the continuous phase (H/DT)
80
Droplet diameter, Dd (µm)
75 Paddle Marine propeller
70 65 60 55 50 45 40 300
143
350
400 450 500 Rotational speed, N (rpm)
550
600
The influence of this parameter was determined using four different rotational speeds (from 300 rpm to 600 rpm) with a paddle impeller of 0.06 m diameter, and the same membranes (5 mm and 50 mm pore sizes) described in Section 3.1. Four different H/DT ratios were studied: 0.7, 0.85, 1.0 and 1.3, that correspond to continuous phase volumes of 1.0, 1.2, 1.4 and 1.8 L, respectively. In this series of trials the paddle impeller with a diameter of 0.06 m was used in a tank of 0.12 m diameter. Results are shown in Fig. 3. No significant difference in droplet size was observed at low rotational speeds. However, an important effect was observed for both membranes when the rotational speed was increased. At low H/DT values and high rotational speeds, larger droplets are obtained with both membranes. The difference in the mean droplet size was 10 mm for the 5 mm pore membrane and 20 mm for the 50 mm pore membrane. Furthermore, the droplet sizes at H/DT ratios of 1.0 and 1.3 are very similar. Hence, H/DT ratios
180
Droplet diameter, Dd (µm)
160 140 120 100 Paddle Marine propeller
80 60 40 300
350
400 450 500 Rotational speed, N (rpm)
550
600
Fig. 2. Influence of impeller rotational speed on mean droplet size of emulsions prepared with metallic membranes and two types of impellers in the emulsification device, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
The plots in Fig. 2 indicate that at low rotational speeds the paddle impeller generates smaller droplets than the marine propeller. This effect is more pronounced for the membrane with pore size of 5 mm. However, at high rotational speeds there was no substantial difference in mean droplet size. Although paddle impellers provide radial flow, the fluid is also pumped upwards and downwards because of the pressure difference between the two sides of the impeller [40]. This axial flow is similar to that produced by the marine propeller. Hence, droplet sizes do not differ appreciably for the two designs. However, the emulsions produced by the marine propeller were characterized by higher span values (0.69 and 0.61 for the 5 mm and 50 mm pore membranes, respectively) than the paddle impeller (0.60 and 0.55 for the same membranes, respectively). Moreover, the emulsions generated using the marine propeller were more sensitive to rotational speed because the difference between the mean droplet size of emulsions prepared at 300 rpm and 600 rpm was larger than for the paddle impeller. This fact makes the paddle impeller easier to control and thus more convenient to use.
Fig. 3. Influence of impeller rotational speed on mean droplet size of emulsions prepared with metallic membranes at different H/DT values in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.06 m. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
144
larger than 1.0 do not lead to a further decrease in droplet size. Therefore, H/DT ¼1.0 can be considered the optimum value, because it is the lowest volume that can yield the smallest droplet size. This H/DT ratio is one of the standard geometric ratios in mixing operations [35,40]. These results can be explained because the smaller the H/DT ratio, the more likely the vortex falls under the impeller level. The practical implication of this situation is that part of the surface of the impeller is not in contact with the liquid. Hence, the entire surface of the paddle is not used in transferring power to the liquid, the torque is lower, and the shear stress decreases. Moreover, the height of continuous phase above the membrane surface decreases as the rotational speed increases, and hence there is an increase in the transmembrane pressure. However, high H/DT ratios imply an increase in liquid volume and a decrease in shear stress according to Eq. (1), which explains the optimum value at H/DT ¼1.0. Similar behaviour was observed for the span values of the two membranes, as shown in Fig. 4. A steady decrease in span was observed as H/DT increased up to 1, but no significant improvement was obtained by increasing this ratio.
tip speeds (u), defined by Eq. (3). The tip speed is related to the tangential velocity. u ¼ pNDi
ð3Þ
Tip speed is a typical scale-up parameter and values for both impellers are indicated in Table 2 and plotted versus droplet size in Fig. 5. Analysis of the data in Fig. 5 indicates that larger droplets were generated with the larger impeller. Although no significant difference was noted for high tip speeds, the influence of the impeller diameter is important at low tip speeds, especially for the 5 mm pore membrane. Because the impeller height is larger for the Di ¼0.09 m impeller, the bottom of the vortex falls under the level of the large impeller even though the tip speed is equivalent. In this situation power is not transferred to the liquid with an associated increase in droplet size. Span values were almost the same for both impellers employed with the 50 mm pore membrane (0.52 and 0.55, respectively), but
80
3.3. Influence of impeller diameter (Di)
0.70 0.68
5 µm 50 µm
0.66
Droplet diameter, Dd (µm)
The influence of impeller diameter on droplet size was studied using the two metallic membranes with paddle impellers (0.06 m and 0.09 m diameter) in a 0.12 m diameter tank. The droplet size was measured at different rotational speeds (N) for both impellers. However, the speed values were chosen according to their
0.64
70 65 60 55 50 45
0.62 Span
Di = 0.06 m Di = 0.09 m
75
40 0.90
0.60
1.10
0.58 0.56
1.70
1.90
150
0.54
Di = 0.06 m Di = 0.09 m
0.50 0.7
1.1
0.9
1.3
H/DT ratio Fig. 4. Influence of H/DT ratio on span values of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.06 m.
Table 2 Equivalent rotational speeds for the same tip speed for impellers of 0.06 m and 0.09 m diameters. Tip speed (m/s)
Rotational speed (rpm)
Rotational speed (rpm)
Di ¼0.06 m
Di ¼ 0.09 m
300 375 450 600
200 250 300 400
Droplet diameter, Dd (µm)
140
0.52
0.94 1.18 1.42 1.88
1.30 1.50 Tip speed, u (m/s)
130 120 110 100 90 80 0.90
1.10
1.30 1.50 Tip speed, u (m/s)
1.70
1.90
Fig. 5. Influence of tip speed on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with paddle impellers of different diameters, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. H/DT ¼ 1.0. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
higher for the larger impeller (0.73) for the trials with the 5 mm pore membrane relative to the small one (0.60).
3.4. Influence of impeller position (Z) The effect of vertical position of the impeller on droplet size distribution was studied as a function of the distance between the impeller and the membrane surface (Z). The study was carried out with both metallic membranes, an H/DT ratio of 1.0, and a paddle impeller (0.06 m diameter) operating at a tip speed of 0.94 m/s (300 rpm). This speed was chosen because at higher values it was not feasible to study a reasonable number of positions because of the vortex effect. Three different Z values were considered: 5 mm (used in previous experiments), 20 mm and 40 mm. Results are shown in Fig. 6. 160
145
No significant effect of the impeller position was observed on the mean droplet size for either membrane at low speeds. Under the chosen conditions no vortex fall takes place. Furthermore, this parameter did not influence the span values. If higher rotational speeds were used and vortex fall would take place, a similar behaviour to Fig. 3 might be expected. However, it should be noted that using a smaller Z implies a longer distance to top surface of the continuous phase. Hence, a wide range of rotational speeds can be used without the vortex falling below the level of the impeller. Consequently, low values of Z would be more suitable for obtaining improved performance in membrane emulsification.
3.5. Influence of impeller-to-tank diameter ratio (Di/DT) The influence of tank size on droplet size was studied with the same paddle impeller (diameter 0.09 m) in three different tanks. Because one experiment required a large volume of continuous
120
150
100
140
80
130
60 40 5 µm
20
50 µm
0 0
20 40 10 30 Distance to membrane surface, Z (mm)
50
Droplet diameter, Dd (µm)
Droplet diameter, Dd (µm)
140
5 µm 50 µm
120 110 100 90 80 70
Fig. 6. Influence of distance between the membrane surface and the impeller (Z) on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m, H/DT ¼ 1.0, N ¼ 300 rpm.
60 0.3
0.7
0.8
145 Droplet diameter, Dd (µm)
95 90 Droplet diameter, Dd (µm)
0.5 0.6 Di/DT ratio
150
100
85 80 75 70 65
140 u = 0.94 m/s u = 1.42 m/s
135 130 125 120 115 110
60
105
55 50 0.10
0.4
0.15
0.20 0.25 Tank diameter, DT (m)
0.30
Fig. 7. Influence of tank diameter on mean droplet size of emulsions prepared with a 5 mm metallic membrane with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼0.09 m, H/DT ¼ 1.0, N ¼ 200 rpm.
100 0.3
0.4
0.5 0.6 Di/DT ratio
0.7
0.8
Fig. 8. Influence of impeller-to-tank diameter ratio on mean droplet size of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. H/DT ¼ 1.0. (A) u¼ 0.94 m/s, (B) 50 mm pore size membrane.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
146
phase, the less favourable conditions reported in Section 3.3 were employed to minimize surfactant consumption. These trials utilized the 5 mm pore membrane, and a tip speed of 0.94 m/s (200 rpm). The three tanks had diameters of 0.12 m, 0.18 m and 0.30 m. These diameters correspond to volumes of 1.4 L, 4.5 L and 25 L, respectively. All were employed at the optimum H/DT ratio (1.0) and results are shown in Fig. 7. The difference in droplet size between the smallest and the largest tank is less than 5 mm. Consequently, an increase in tank size does not lead to worse performance. The smallest droplet size corresponds to emulsions prepared in a 0.18 m diameter tank. For this impeller/tank combination Di/DT ¼0.5. No influence was observed in the span values (0.65 70.05). The experiments represented in Fig. 7 correspond to three different impeller-to-tank diameter ratios, i.e., 0.33, 0.5 and 0.75. The experiments in Fig. 8 correspond to both membranes, using the same impeller-to-tank diameter ratios and a tip speed of 0.94 m/s (see Fig. 8A). In addition, another tip speed (1.42 m/s) was employed in trials with the membrane of 50 mm pore size (see Fig. 8B). The mean droplet size of emulsions obtained at a tip speed of 0.94 m/s varied by only a few percent. The maximum value was obtained at Di/DT ratio of 0.75. However, different behaviour was observed for the trials at a tip speed of 1.42 m/s (Fig. 8B). The mean droplet size was significantly smaller for low Di/DT ratios. This result might be explained by the vortex fall at high impeller rotational speed, so that part of the impeller does not transfer power to the liquid. This vortex fall did not take place at a tip speed of 0.94 m/s. In addition, it has been reported [40] that the higher the Di/DT ratio the lower the average shear rate, so that the droplet size should increase. The standard ratio in mixing operations is Di/DT ¼ 0.3; however, the optimum ratio for membrane emulsification depends on the type of impeller and operating conditions. The span values were relatively invariant at values between 0.50 and 0.55.
3.6. Influence of baffles Vortex formation was observed in experiments at high rotational speeds. Vortexing was a problem, because power was not completely transferred to the continuous phase. It has been reported [35,41] that the vortex is caused by the tangential component of the continuous phase velocity. This factor is also related to droplet detachment [3]. To determine the influence of tangential velocity and vortex, experiments were performed in a tank with and without baffles using both metallic membranes (5 mm and 50 mm pores) with the 0.06 m diameter impeller at a rotational speed of 600 rpm. The baffles had a height of 0.17 m and a width of 0.014 m. Results are shown in Fig. 9, for several emulsification times, t. It is observed in Fig. 9A that the differences in the droplet size distribution are not very significant for the 5 mm pore size membrane and that the droplet size is slightly larger in the baffled tank. However, the droplet size distribution obtained after 30 min with the 50 mm pore size membrane was significantly different. The droplets were smaller and the size distribution was wider in the baffled tank. The evolution of the emulsion with time was studied in the baffled tank. Breakup of droplets occurred for the 50 mm membrane, as shown in Fig. 9B, but it was not noticed for the 5 mm membrane (Fig. 9C). Moreover, no breakup was observed in the unbaffled tank (Fig. 9D). This fact could be explained because of the large size of the droplets obtained with the 50 mm membrane. Such droplets are more sensitive to shear in the vicinity of the baffles [35], yielding a wider droplet size distribution. 3.7. Membrane emulsification in large-volume tanks The purpose of scale-up of membrane emulsification is to obtain emulsions at large scale having the same droplet size, but
30
30
20 15
25
5 µm unbaffled 5 µm baffled 50 µm unbaffled 50 µm baffled
Volume (%)
Volume (%)
25
10 5 0 0.01
0.1
1 10 Diameter (µm)
100
15
0.1
1 10 Diameter (µm)
100
1000
100
1000
30 Baffled t = 10 min Baffled t = 20 min Baffled t = 30 min Unbaffled t = 30 min
25
10 5 0 0.01
10
0 0.01
1000
Volume (%)
Volume (%)
20
15
5
30 25
20
Baffled t = 10 min Baffled t = 20 min Baffled t = 30 min Unbaffled t = 30 min
t = 30 min t = 45 min t = 60 min
20 15 10 5
0.1
1 10 Diameter (µm)
100
1000
0 0.01
0.1
1 10 Diameter (µm)
Fig. 9. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di ¼ 0.06 m, DT ¼0.15 m, H/DT ¼1.0, N ¼ 600 rpm. (A) t¼ 30 min, (B) 50 mm pore size membrane, (C) 5 mm pore size membrane (D) 50 mm pore size membrane and unbaffled tank.
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
Table 3 Scale-up criteria for stirred tanks. Scale-up parameter
Equation
Tip speed
u ¼ pNDi
Reynolds number
Re ¼ P ¼ V
Power per unit volume
Conditions in the larger tank ð4Þ
rc NDi 2 ð5Þ mc 2pNM V
Di1 Di2 Di1 2 N2 ¼ N1 Di2 3 Di2 P2 ¼ P 1 Di1 N2 ¼ N1
ð6Þ
Table 4 Conditions for scaling-up membrane emulsification using paddle impellers (0.06 and 0.09 m diameter) in stirred tanks (1.4 L and 4.5 L continuous phase) keeping Di/DT ¼ 0.5. Parameter
Small tank (1.4 L continuous phase)
Large tank (4.5 L continuous phase)
Tip speed Reynolds number Power per unit volume
N1 ¼ 300 rpm N1 ¼ 300 rpm P1 ¼ 1.08 W (N1 ¼ 300 rpm)
N2 ¼ 200 rpm N2 ¼ 133 rpm P2 ¼3.65 W (N2 4400 rpm)
ð7Þ ð8Þ
147
ð9Þ
35 30 25
V = 1.4L; u = 1.42 m/s V = 4.5L; u = 1.42 m/s
30
V = 4.5L V = 25L
25
20
Volume (%)
Volume (%)
35
V = 1.4L; u = 0.94 m/s V = 4.5L; u = 0.94 m/s
15
20 15
10 10
5 5
0 0.1
1
10 Diameter (µm)
100
1000
0 0.1
1
10 Diameter (µm)
100
1000
40 35
Fig. 11. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di/DT ¼ 0.33, H/DT ¼ 1.0, u¼ 0.94 m/s.
V = 1.4L; u = 0.94 m/s V = 4.5L; u = 0.94 m/s
Volume (%)
30 25
V = 1.4L; u = 1.42 m/s V = 4.5L; u = 1.42 m/s
20 15 10 5 0 0.1
1
10 Diameter (µm)
100
1000
Fig. 10. Droplet size distributions of emulsions prepared with metallic membranes in the emulsification device with a paddle impeller, using as continuous phase a Tween 20 2 wt% with CMCNa 0.01 wt% solution and extra virgin olive oil as dispersed phase. Di/DT ¼ 0.5, H/DT ¼ 1.0. (A) 5 mm pore size membrane, (B) 50 mm pore size membrane.
maintaining at least geometric similarity. Different criteria for scaling-up involving stirred tank mixing have been considered and the equations (Eq. (4)–(9)) are summarized in Table 3, in which the subscript ‘‘1’’ refers to the laboratory (small) unit and ‘‘2’’ to the larger scale unit. Experiments were conducted in two stirred tanks with diameters of 0.12 m and 0.18 m. These diameters correspond to continuous phase volumes of 1.4 L and 4.5 L, respectively. Two paddle impellers with diameters of 0.06 m and 0.09 m were
utilized, H/DT and Di/DT ratios were set to 1.0 and 0.5, respectively, and two impeller tip speeds were used (0.94 m/s and 1.42 m/s) along with flat metallic membranes with pore sizes of 5 mm and 50 mm. The scale-up parameters in Table 3 were explored in the experimental trials. Tip speed was the first criterion used and results are shown in Fig. 10. The droplet size distributions of emulsions prepared at the same tip speed in different tanks, maintaining geometric similarity, are the same. Data were obtained at two different tip speeds and for the two membranes with different pore sizes. The conditions for using P/V and Reynolds number as scale-up criteria are summarized in Table 4. The volumetric power input, P/V, commonly used to scale-up mixing in stirred tanks, does not seem to be a suitable criterion for scale-up of membrane emulsification because it implies using very high rotational speeds. The entries in Table 4 for scaling-up experiments at 300 rpm in the small tank would need using rotational speeds higher than 400 rpm in the large tank. However, even at a rotational speed of 300 rpm, a different droplet size is produced. Neither is the Reynolds number a good criterion for scale-up, because the rotational speed for the large tank is underestimated. In order to verify the suitability of tip speed as scale-up criterion, which is considered as non-conservative for mixing operations, another experiment was carried out in a larger tank with 25 L of continuous phase (Di ¼ 0.09 m, Di/DT ¼0.33 and H/DT ¼ 1.0). The results were compared with those obtained in the tank with 4.5 L of continuous phase. The 5 mm pore size
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
148
membrane was used and rotational speed was set to 300 rpm in the smaller tank. Droplet size distributions are shown in Fig. 11. If geometric ratios and tip speed are kept constant a significant increase in tank size does not affect droplet size. Even though tip speed is not considered the best scale-up parameter in mixing operations, it seems to be appropriate for membrane emulsification. This criterion could also be expressed in terms of dimensionless numbers containing tip speed, such as the capillary and Euler numbers [34]. However, diffusion, which has been reported to affect membrane emulsification [42], does not seem to have much influence under the experimental conditions and for the emulsion used in this work. If the process were affected by diffusion, other scale-up parameters related to mass transfer (e.g., the mass transfer coefficient) [41] would be more suitable. It has been also reported that tip speed is related to the maximum shear rate [40], which could mean that the maximum shear stress is the key parameter in determining droplet size during membrane emulsification. It should also be noted that the membrane active area in the emulsification unit shown in Fig. 1 is placed within the forced vortex region, where shear stress is not affected by the tank walls [8,35,41]. Thus, droplet detachment might be different if the membrane active area were increased.
D90 D[4,3] Dd Di Di/DT DT H H/DT M n N P P/V Re t u V Z
Diameter for which 90% of distribution volume has smaller size (mm) Diameter of the sphere with volume equivalent to that of the droplet (mm) Mean droplet diameter (mm) Impeller diameter (m) Impeller-to-tank diameter ratio Tank diameter (m) Height of liquid (continuous phase) in the tank (m) Continuous phase height-to-tank diameter ratio Torque (N m) Refractive index Rotational speed (rps) Power input (W) Power per unit volume (W/m3) Reynolds number for stirring time (s) Impeller tip speed (m/s) Volume of liquid (continuous phase) in the tank (m3) Distance from impeller to membrane surface (m)
Greek letters 4. Conclusions Membrane emulsification experiments to produce O/W emulsions, reported in this work, lead to the following conclusions:
Paddle impellers exhibit better performance than marine
propellers for membrane emulsification because the resulting emulsions are less sensitive to rotational speed. However, differences are higher at low rotational speeds. Low H/DT ratios lead to large droplets, especially for high rotational speeds. The optimum value was H/DT ¼1.0, that corresponds to standard ratios in mixing operations. Furthermore, the Di/DT ratio has little influence on droplet size, nor do the diameters of the impeller and the tank. The distance between the impeller and the membrane surface does not affect the droplet size of the emulsion at low speeds. However, when baffles are used droplets are larger than in an unbaffled tank. In addition, breakup of the droplets was observed for large droplets in baffled tanks. Impeller tip speed is a key parameter for scaling-up, because at a certain value the same droplet size distributions were obtained, when geometric similarity was maintained. Finally, optimum conditions in mixing operations are not necessarily the optimum conditions for membrane emulsification.
Acknowledgements This work was supported by the MICINN, Spain, under the grants CTQ2007-65348 and MICINN-10-CTQ2010-20009-C02-01. M.A. Sua´rez acknowledges receipt of a graduate fellowship from FPU Program (MECD, Spain), co-funded by European Social Fund. The authors thank Prof. Charles Hill for fruitful discussions.
Nomenclature D10 D50
Diameter for which 10% of distribution volume has smaller size (mm) Diameter for which 50% of distribution volume has smaller size (mm)
g mc md rc rd t
Shear rate (s 1) Continuous phase dynamic viscosity (Pa s) Dispersed phase dynamic viscosity (Pa s) Continuous phase density (kg/m3) Dispersed phase density (kg/m3) Shear stress (Pa)
Subscripts 1 2
Small tank Large tank
References [1] T. Nakashima, M. Shimizu, M. Kukizaki, Membrane emulsification by microporous glass, Key Eng. Mater. 61&62 (1991) 513–516. [2] T. Nakashima, M. Shimizu, M. Kukizaki, Particle control of emulsion by membrane emulsification and its applications, Adv. Drug Delivery Rev. 45 (2000) 47–56. [3] S.J. Peng, R.A. Williams, Controlled production of emulsions using a crossflow membrane: Part I: Droplet formation from a single pore, Chem. Eng. Res. Des. 76 (1998) 894–901. ¨ [4] V. Schroder, O. Behrend, H. Schubert, Effect of dynamic interfacial tension on the emulsification process using microporous ceramic membranes, J. Colloid Interface Sci. 202 (1998) 334–340. ¨ [5] V. Schroder, H. Schubert, Production of emulsions using microporous ceramic membranes, Colloids Surf., A: Physicochem. Eng. Aspects 152 (1999) 103–109. [6] C. Charcosset, H. Fessi, Preparation of nanoparticles with a membrane contactor, J. Membr. Sci. 266 (2006) 115–120. [7] N. Sheibat-Othman, T. Burne, C. Charcosset, H. Fessi, Preparation of pHsensitive particles by membrane contactor, Colloids Surf., A: Physicochem. Eng. Aspects 315 (2008) 13–22. [8] S.R. Kosvintsev, G. Gasparini, R.G. Holdich, I.W. Cumming, M.T. Stillwell, Liquid–liquid membrane dispersion in a stirred cell with and without controlled shear, Ind. Eng. Chem. Res. 44 (2005) 9323–9330. [9] M.T. Stillwell, R.G. Holdich, S.R. Kosvintsev, G. Gasparini, I.W. Cumming, Stirred cell membrane emulsification and factors influencing dispersion drop size and uniformity, Ind. Eng. Chem. Res. 46 (2007) 965–972. [10] M.M. Dragosavac, M.N. Sovilj, S.R. Kosvintsev, R.G. Holdich, G.T. Vladisavljevic, Controlled production of oil-in-water emulsions containing unrefined pumpkin seed oil using stirred cell membrane emulsification, J. Membr. Sci. 322 (2008) 178–188. [11] E. Egidi, G. Gasparini, R.G. Holdich, G.T. Vladisavljevic, S. Kosvintsev, Membrane emulsification using membranes of regular pore spacing: Droplet size
´rez et al. / Journal of Membrane Science 430 (2013) 140–149 M.A. Sua
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22] [23]
[24]
[25] [26]
and uniformity in the presence of surface shear, J. Membr. Sci. 323 (2008) 414–420. I. Kobayashi, M. Yasuno, S. Iwamoto, A. Shono, K. Satoh, M. Nakajima, Microscopic observation of emulsion droplet formation from a polycarbonate membrane, Colloids Surf., A: Physicochem. Eng. Aspects 207 (2002) 185–196. ¨ A. Trentin, C. Guell, F. Lo´pez, M. Ferrando, Microfiltration membranes to produce BSA-stabilized emulsions by premix membrane emulsification, J. Membr. Sci. 356 (2010) 22–32. P.J. Dowding, J.W. Goodwin, B. Vincent, Production of porous suspension polymer beads with a narrow size distribution using a cross-flow membrane and a continuous tubular reactor, Colloids Surf., A: Physicochem. Eng. Aspects 180 (2001) 301–309. M.J. Geerken, R.G.H. Lammertink, M. Wessling, Interfacial aspects of water drop formation at micro-engineered orifices, J. Colloid Interface Sci. 312 (2007) 460–469. M.J. Geerken, M.N.W. Groenendijk, R.G.H. Lammertink, M. Wessling, Microfabricated metal nozzle plates used for water-in-oil and oil-in-water emulsification, J. Membr. Sci. 310 (2008) 374–383. V. Schadler, E.J. Windhab, Continuous membrane emulsification by using a membrane system with controlled pore distance, Desalination 189 (2006) 130–135. G.T. Vladisavljevic, R.A. Williams, Manufacture of large uniform droplets using rotating membrane emulsification, J. Colloid Interface Sci. 299 (2006) 396–402. R.G. Holdich, M.M. Dragosavac, G.T. Vladisavljevic, S.T. Kosvintsev, Membrane emulsification with oscillating and stationary membranes, Ind. Eng. Chem. Res. 49 (2010) 3810–3817. A.J. Gijsbertsen-Abrahamse, A. van der Padt, R.M. Boom, Status of cross-flow membrane emulsification and outlook for industrial application, J. Membr. Sci. 230 (2004) 149–159. D.X. Hao, F.L. Gong, G.H. Hu, Y.J. Zhao, G.P. Lian, G.H. Ma, Z. Su, Controlling factors on droplets uniformity in membrane emulsification: experiment and modeling analysis, Ind. Eng. Chem. Res. 47 (2008) 6418–6425. M.A. Suarez, J. Coca, C. Pazos, Membrane emulsification: factors influencing the size and distribution of drops, Ing. Quim. 505 (2012) 58–73. G.T. Vladisavljevic, H. Schubert, Influence of process parameters on droplet size distribution in SPG membrane emulsification and stability of prepared emulsion droplets, J. Membr. Sci. 225 (2003) 15–23. G.T. Vladisavljevic, U. Lambrich, M. Nakajima, H. Schubert, Production of O/W emulsions using SPG membranes, ceramic–aluminium oxide membranes, microfluidizer and a silicon microchannel plate-a comparative study, Colloids Surf., A: Physicochem. Eng. Aspects 232 (2004) 199–207. ¨ ard, ˚ S.M. Joscelyne, G. Trag Membrane emulsification—a literature review, J. Membr. Sci. 169 (2000) 107–117. G.T. Vladisavljevic, R.A. Williams, Recent developments in manufacturing emulsions and particulate products using membranes, Adv. Colloid Interface Sci. 113 (2005) 1–20.
149
[27] M. Kukizaki, Shirasu porous glass (SPG) membrane emulsification in the absence of shear flow at the membrane surface: influence of surfactant type and concentration, viscosities of dispersed and continuous phases, and transmembrane pressure, J. Membr. Sci. 327 (2009) 234–243. [28] G. Vladisavljevic, M. Shimizu, T. Nakashima, Production of multiple emulsions for drug delivery systems by repeated SPG membrane homogenization: Influence of mean pore size, interfacial tension and continuous phase viscosity, J. Membr. Sci. 284 (2006) 373–383. [29] R.A. Williams, S.J. Peng, D.A. Wheeler, N.C. Morley, D. Taylor, M. Whalley, D.W. Houldsworth, Controlled production of emulsions using a crossflow membrane. Part II: Industrial scale manufacture, Chem. Eng. Res. Des. 76 (1998) 902–910. [30] I. Limayem Blouza, C. Charcosset, S. Sfar, H. Fessi, Preparation and characterization of spironolactone-loaded nanocapsules for paediatric use, Int. J. Pharm. 325 (2006) 124–131. [31] Q. Yuan, R.A. Williams, Large scale manufacture of magnetic polymer particles using membranes and microfluidic devices, China Part 5 (2007) 26–42. [32] E. Lepercq-Bost, M.L. Giorgi, A. Isambert, C. Arnaud, Use of the capillary number for the prediction of droplet size in membrane emulsification, J. Membr. Sci. 314 (2008) 76–89. ¨ R.M. Boom, Parallelized edge-based [33] K.C. v. Dijke, G. Veldhuis, K. Schroen, droplet generation devices, Lab Chip 9 (2009) 2824–2830. [34] M.A. Suarez, G. Gutierrez, J. Coca, C. Pazos, Stirred tank membrane emulsification using flat metallic membranes: a dimensional analysis, Chem. Eng. Process, Unpublished results. [35] F.A. Holland, F.S. Chapman, Liquid Mixing and Processing in Stirred Tanks, Reinhold Publishing, New York, 1966. [36] S. Nagata, Mixing. Principles and Applications, John Wiley & Sons Inc., New York, 1975. [37] M. Ciofalo, Turbulent flow in closed and free surface unbaffled tanks stirred by radial impellers, Chem. Eng. Sci. 51 (14) (1996) 3557–3573. [38] H.J. Henzler, J. Kauling, Scale up of mass transfer in highly viscous liquids, in: ¨ European Conference on Mixing. Wurzburg, Germany, Cranfield BHRA, 1985, pp. 303–312. [39] J.A. Sa´nchez Pe´rez, E.M. Rodrı´guez Porcel, J.L. Casas Lo´pez, J.M. Ferna´ndez Sevilla, Y. Chisti, Shear rate in stirred tank and bubble column bioreactors, Chem. Eng. J. 124 (2006) 1–5. [40] J.Y. Oldshue, Fluid Mixing Technology, McGraw Hill Publications Co, New York, 1983. [41] Z. Sterbacek, P. Tausk, Mixing in the Chemical Industry, Pergamon Press, Oxford, 1965. ¨ ard, ˚ ¨ ard, ˚ [42] M. Rayner, G. Trag C. Trag The impact of mass transfer and interfacial tension expansion rate on droplet size in membrane emulsification processes, Colloids Surf., A: Physicochem. Eng. Aspects 266 (2005) 1–17.