Maryam
Parhizkar
a,
Eleanor
Stride
ab and
Mohan
Edirisinghe
*a
aDepartment of Mechanical Engineering, University College London, Torrington Place, London, WC1E 7JE, UK. E-mail: m.edirisinghe@ucl.ac.uk; Fax: +44 (0)2073880180; Tel: +44 (0)2076793942
bInstitute of Biomedical Engineering, Department of Engineering Science, University of Oxford, Old Road Campus Research Building, Headington, Oxford OX3 7DQ, UK
First published on 30th April 2014
This work investigates the generation of monodisperse microbubbles using a microfluidic setup combined with electrohydrodynamic processing. A basic T-junction microfluidic device was modified by applying an electrical potential difference across the outlet channel. A model glycerol air system was selected for the experiments. In order to investigate the influence of the electric field strength on bubble formation, the applied voltage was increased systematically up to 21 kV. The effect of solution viscosity and electrical conductivity was also investigated. It was found that with increasing electrical potential difference, the size of the microbubbles reduced to ~25% of the capillary diameter whilst their size distribution remained narrow (polydispersity index ~1%). A critical value of 12 kV was found above which no further significant reduction in the size of the microbubbles was observed. The findings suggest that the size of the bubbles formed in the T-junction (i.e. in the absence of the electric field) is strongly influenced by the viscosity of the solution. The eventual size of bubbles produced by the composite device, however, was only weakly dependent upon viscosity. Further experiments, in which the solution electrical conductivity was varied by the addition of a salt indicated that this had a much stronger influence upon bubble size.
Fluid flow in the channels of microfluidic devices has most commonly been controlled using high precision mechanical pumps.14 However, another type of flow in microchannels, broadly refrered to as electroosmotic flow,15 initiated by the application of an electric field, has also been studied extensively.16 This method of driving and controlling the operating fluid, has some distinct advantages due to the localization of the electrical forces in these miniaturized devices. High electric fields can be obtained with relative ease and they can assist with the flow of fluids in the microchannels.17 Thus, in order to alleviate the difficulties of excessive pressure gradients associated with microfluidic pumps in microchannels, pressure driven flows are often replaced by electroosmotic flows.18
Electrically driven microfluidic devices have previously been used in many studies for mixing of two phase flows,19 and generation of monodisperse droplets,20–22 fibers23,24 and microbubbles.25 Kim et al.20 developed a microchip droplet generator using an electrohydrodynamic actuation method. Droplet formation was controlled by the application of an electric field between the charged liquid sample and a ground electrode without the need for an external pneumatic pump. Srivastava and coworkers24 described a microfluidic based electrospinning method to fabricate hollow and core/sheath nanofibers. Of particular relevance to the present study, Pancholi et al.25 produced phospholipid coated microbubbles with diameters smaller than 8 μm using a device consisting of a combined T-junction microfluidic and electrospraying device. However, the size distibution of microbubbles produced in this study was still relatively broad.
Using a T-junction microfluidic device is one of the easiest methods of producing highly monodisperse microbubbles. However, to generate bubbles with diameters smaller than the geometrical diameter of the channels, using mechanically assembled devices is challenging due to constraints on capillary size, especially at higher viscosities.26 As indicated above, the use of an electric field can offer significant advantages for liquid manipulation in microchannels. The small channel cross sectional area presents high electrical resistance to ionic currents, which allows high electric fields to be maintained with low currents and hence provides control over the liquid velocity.27 This in turn provides control over the breakup of the gas column and formation of bubbles. While microfluidics and EHD have been separately used to produce microbubbles, to the best of the authors' knowledge, the direct combination of these two methods to form monodisperse bubbles has not been reported in the literature. The capillary embedded T-junction device described in this work provides a simple but yet robust means of producing highly monodisperse microbubbles. However, because the channel diameters are relatively large (compared with e.g. devices prepared via photolithography) the production of bubbles that are signfiicantly smaller than the channel diameter is not viable with purely mechanically driven flow. In this work, we present a microfluidic system with integrated electrohydrodynamic focusing with the aim of both reducing bubble size and maintaining monodispersity. We investigate the effect of applied voltage, solution viscosity and electrical conductivity on the production of microbubbles and their characteristics. We show that by introducing an electric field directly into the bubble break up region, the flow of the continuous phase is assisted by electrohydrodynamic forces and bubbles with almost an order of magnitude smaller than the channel diameter can be generated.
Aqueous solution | Viscosity (μ/mPa s) | Surface tension (σ/mN m−1) | Electrical conductivity (k/μS cm−1) | pH |
---|---|---|---|---|
5 wt.% glycerol, 1 wt.% SLS | 1.3 | 50 | 120 | 4.8 |
50 wt.% glycerol, 1 wt.% SLS | 6 | 56 | 18 | 8.2 |
50 wt.% glycerol, 1 wt.% SLS, 1 wt.% NaCl | 6 | 56 | 1500 | 7.4 |
65 wt.% glycerol, 1 wt.% SLS | 15 | 57 | 16 | 8.3 |
75 wt.% glycerol, 1 wt.% SLS | 36 | 59 | 14 | 8.4 |
Fig. 1 T-junction setup a) and schematic of the T-junction setup and bubble formation without b) and with c) an applied electric field. |
Rel = ρlUl/μl |
Ca = μlUl/σl |
Where, Rel and Ca refer to the Reynolds and Capillary number and μl, Ul, σl and ρl, are the liquid viscosity, velocity, surface tension and density, respectively. Re represents the ratio of the inertial to viscous forces and Ca represents the ratio of viscous forces and surface tension acting on an interface. For the liquid phase 7 × 10−5 ≤ Rel ≤ 9.3 × 10−3. When Rel < 1 flow is dominated by viscous stresses and pressure gradients and therefore inertial effects are negligible. According to electrohydrodynamic theory,28,29 the electric Korteweg–Helmholtz force exerted per unit volume of fluid can be written as:
(1) |
Where ρe is the volume charge density, ρ is the liquid density, ε is the dielectric constant for the liquid and Ē is the electric field strength. The first term on the right hand side of eqn (1) represents the Coulomb force acting on the free charge and can be neglected when the current is small. The second term is the dielectrophoresis force exerted on the liquid due to the spatial gradient in the permittivity,30 which is classified as the main force acting on the liquid–gas interface. For a spherical bubble:
(2) |
Where Db is the bubble diameter. Since the dielectric constant of a gas is smaller than that of a liquid, FDEP at the liquid–gas interface will act towards the centre of the bubble. In the presence of the electric field, a bubble emerging into the outlet channel becomes polarized and when the electric field is applied, the bubble moves away from the contact area. By varying the pressure distribution in the liquid phase, this force increases the elongation of the gas column into the outlet channel. The third term in eqn (1) refers to the electrostriction force which is negligible in this case due to the minimal influence of fluid compressibility on bubble formation.
In the absence of an electric field, the competition between liquid and gas pressure, viscous forces and interfacial tension controls the breakup of the gas column into bubbles. Once the sum of the viscous stress and pressure difference due to the obstruction of channel by the emerging gas column exceeds the capillary pressure, detachment begins. The capillary force Fσ is given by the difference between the Laplace pressures upstream and downstream of the emerging bubble multiplied by the projected area of the emerging interface (where R1 and R2 are the radii of curvature axially and radially and Ainterface is the projected area).
(3) |
The viscous shear force Fτ is given by the product of viscous stress acting on the emerging interface and the projected area of the emerging interface.
Fτ ≈ μlQlAinterface | (4) |
Finally, following Garstecki et al.31 the squeezing pressure force:
Fp ≈ ΔPcAinterface | (5) |
When an electric field is applied, electrical charge accumulates at the gas liquid interface which behaves as a capacitor. As the voltage increases, the charge build up at the interface increases resulting in an additional force on the gas column. The resulting elongation of the gas column in the axial direction and radial compression accelerates the breakup process and therefore leads to smaller bubbles.
Fig. 2 High speed camera images of microbubbles at the tip of the outlet for 50 wt% glycerol solution at applied voltages of 0, 6 and 12 kV. Scale bar is 1.6 mm. |
As well this observable effect at the tip of the outlet channel, as above, a tangential electrical force is created that leads to faster breakup of the gas column at the junction, therefore reducing the detachment time and leading to the formation of smaller bubbles at a faster rate.34
Once the bubbles were formed at the minimum gas pressure (Pgmin) for each solution, the applied voltage was increased to 6 kV, where the meniscus at the tip of the outlet channel became thinner while the jet diameter was reduced and therefore microbubble size decreased. For instance, for the lowest solution viscosity (1.3 mPa s) the diameter of microbubbles produced without an electric field was 170 μm, which reduced to 120 μm at 6 kV (Fig. 3). By increasing the voltage to 9 kV, a cone jet was created at the tip of the outlet channel and the size of bubbles reduced further, to 40 μm with a polydispersity index (PDI, defined as the ratio between the standard deviation and mean diameter in percentage) of ~1%. At 12 kV, the cone jet broke up into a spray of fine liquid threads and bubbles with even smaller diameters (30 ± 0.95 μm) were produced.
This process was repeated for solutions with higher glycerol concentration and viscosities of 6, 15 and 36 mPa s and it was shown that increasing the voltage also affected bubble size for the highest viscosity solution while the smallest microbubble diameter of 25 μm was produced for the solution with 36 mPa s viscosity at 12 kV.
Increasing the applied voltage to 15 kV, did not change the bubble size significantly at any of the glycerol concentrations. However bubble stability decreased, most likely as a result of coalescence due to the higher surface charge. Fig. 4 shows microscopic images of microbubbles produced from 75% glycerol solution at constant liquid flow rate and applied voltages of 12–21 kV. It is evident from the images that at 12 and 15 kV the bubble size was the same (25 μm) and they were near monodisperse.
Fig. 4 Optical micrographs of microbubbles from a solution with 75% glycerol concentration at constant liquid flow rate of 0.01 ml min−1 at applied voltages of a) 12, b) 15 and c) 21 kV. |
However, increasing the voltage supply to 21 kV led to a much broader size distribution. This suggests that the optimum voltage for this system is 12 kV and increasing the voltage above this rate only reduces microbubble stability and monodispersity. A series of graphs were plotted (Fig. 5) to show the variation in microbubble size with increasing voltage. In all cases, as the voltage increased bubble diameter decreased; however a dramatic decrease in bubble diameter was observed between 6–9 kV for the solution with lowest viscosity that was not seen in the other solutions (Fig. 5a). This is most likely to be due to the fact that the solution with 5% glycerol concentration has a much higher dielectric constant35 and therefore the effect of applied voltage on bubble diameter is greater. The scaling law proposed by Pantano et al.36 predicts that the diameter of droplets produced by electrospraying is inversely proportional to the liquid dielectric constant.
Fig. 5 a) Graph showing variation of bubble diameter with applied voltage for solution viscosities of 1.3, 6, 15 and 36 mPa s, and b) 3D plot of dimensionless bubble diameter with respect to voltage and capillary number increment. Error bars in Fig. 5a refer to repeat experiments. |
In order to investigate the effect of viscosity, surface tension, flow rate ratio and applied voltage supply in parallel, a 3D plot of the variation of ratio of bubble diameter to channel width was plotted as shown in Fig. 5b. It can be observed that for each value of the capillary number, with increasing voltage the bubble to channel diameter ratio decreased dramatically between 0 and 9 kV. The reduction in this value is less significant, however at larger voltages. According to Ku and Kim,37 for highly conducting and viscous liquids, the size of droplets electrosprayed from a “Taylor” cone are found to be relatively insensitive to the applied voltage and as long as the corona discharge density is not too high, monodisperse droplets are produced. Corona discharge is caused by the ionization of the surrounding medium that occurs once the electric field strength exceeds a certain level (the corona threshold voltage) while conditions are inadequate for a complete electrical breakdown. Above this voltage, there is a limited region, in which current increases proportionately with voltage according to Ohm's law. After this region, the current increases more rapidly, leading to complete breakdown and arcing or sparking at a point called the breakdown potential. It is also shown in Fig. 5b that with increasing capillary number, there is a smaller reduction in bubble size for the same increase in applied voltage. This suggested that there other parameters such as solution electrical conductivity and relative permittivity as mentioned previously that influence the bubble formation process. To investigate this further, NaCl was added to the solution keeping the concentration of glycerol constant at 50% in order to increase the electrical conductivity while keeping the other solution parameters constant. The results are plotted in Fig. 6 and as predicted, by increasing the electrical conductivity of the solution while keeping the viscosity and flow ratio constant, the influence of the voltage supply on microbubble size increases. This explains the dramatic decrease in bubble diameter in the graph representing the solution with lowest concentration of glycerol compared with the other graphs in Fig. 5. The electrical conductivity of the liquid phase is one of the key parameters in determining and predicting the bubble diameter.
Fig. 7 Variation in bubble dimensionless diameter for solutions with a viscosity of a) 1.3, b) 6,) c) 15 and d) 36 mPa s. (PDI < 1%). |
By further increasing the applied voltage, the width of the neck during the breakup reduces. However, similar to observations reported by Kim et al.38 jetting occurs and there is very little further effect on the bubble size. The production of monodisperse bubbles was found to cease at voltages more than 20 kV. For each case, an asymptotic curve is fitted to show this limit in the reduction of bubble size. It can also be seen that the solutions of higher concentration of glycerol (i.e. higher viscosity) follow a similar trend, whilst in the case for the solution with the lowest viscosity the trend is slightly changed due to the fact that the electrical conductivity is comparatively lower. For the range of Ca numbers investigated, a general predictive model is obtained where the normalized bubble diameter can be estimated as the following:
(6) |
This model can predict the dimensionless bubble size for a range of capillary numbers 0.001 ≤ Ca ≤ 0.04, with approximately 8% error. In Fig. 8, the experimental values for Db/Dch are plotted against the predicted values and the proximity of the experimental data to the parity line suggests that the predictive model is in agreement with the experimental data especially for the obtained values of Db/Dch < 0.6. This model does not take into account the geometrical aspects of the channel (i.e. the gap between the capillaries), as these parameters also affect the bubble size.
Fig. 8 Graph representing experimental data for dimensionless bubble diameter against predicted values. |
Fig. 9 depicts the number of bubbles produced with and without the presence of an electric field for different solution viscosities over a fixed time period of 5 s and collection area of 1.5 mm2. It is shown that by increasing the voltage not only did the bubbles become smaller but also the production rate increased. In addition, from the data obtained from the high speed camera images, for a given liquid flow rate of 0.01 ml min−1, the number of bubbles in every 1 ml of the collected sample is between 2 × 106 and 6 × 106 depending on the bubble size.
A key question is whether or not the bubble size can be further reduced using this technique to enable production of microbubbles with diameters <10 μm (such as would be required for intravenous administration in biomedical applications). The most obvious means of achieving this would be to use a device with smaller capillaries. If similar ratios of bubble: capillary diameter could be achieved as shown in Fig. 7 then using capillaries of half the size (50 μm internal diameter) would yield microbubbles in the desired range, whilst still giving greatly reduced risk of clogging compared with existing devices in which the bubble:channel diameter ratio is ~1. This raises the further question as to whether or not a proportional reduction in bubble size can be achieved in smaller capillaries with the application of an electric field. Eqn (2) indicates that the electrophoretic force decreases with bubble volume whilst eqn (3) shows that the capillary force scales approximately with radius. Eqn (4) and (5) indicate that the viscous and pressure forces scale with projected area but also with liquid volume flow rate and gas pressure respectively. The effect on bubble size will therefore depend upon the relative change in these latter two quantities. Previous work by the authors26 has shown that the rate at which bubble size decreases with the ratio of liquid to gas flow rates increases with decreasing capillary size. This would suggest that even with a reduction in the relative influence of FDEP it would still be possible to generate microbubbles substantially smaller than the capillary. Future work will aim to demonstrate this.
There are numerous applications from biomedical imaging to food and water treatment that require microbubbles with a controlled range of sizes. For instance, microbubble induced cavitation has been proposed as an innovative method for minimally-invasive drug delivery.10 Microbubble assisted flotation is widely used in the recovery of fine mineral particles and flotation and for solid–liquid separation to remove pollutants. Recent bench studies of flotation of different minerals; with injection of microbubbles (40 μm, mean diameter) to lab cells (in addition to the cell generated coarse bubbles) have significantly improved separation efficiency when compared to the mill standard.41 Microbubbles have been shown to increase fermentation rates in the production of biofuels whilst mixtures of ozone nanobubbles with oxygen microbubbles have been shown to be more effective in fighting bacteria than conventional ozone saturated water.42 Each of these applications required the concentration, size and size distribution of microbubbles to be tightly controlled in order to maximise efficacy. Many of the key characteristics of microbubbles are directly related to their size (e.g. stability, buoyancy and surface activity) and in this work we have presented a new technique for microbubble formation which offers excellent control over bubble size and polydispersity, continuous production with lower risk of clogging and the potential for multiplexing to achieve higher production rates. However, in order to be useful in biomedical applications, the size of the microbubbles need to be further reduced to <10 μm and this is being addressed in our current work as indicated above.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4lc00328d |
This journal is © The Royal Society of Chemistry 2014 |