Ultra-thin liquid film extraction based on a gas–liquid–liquid double emulsion in a microchannel device

Abstract

A new microfluidic extraction method using a gas–liquid–liquid double emulsion with ultra-thin solvent film as the working system was developed, overcoming the defect of low extraction efficiency at extreme phase ratios (R > 10) by significantly reducing the mass transfer distance. Interesting bubble coalescence in the emulsion generation was reported and laws of bubble diameter (0.73–1.08 mm) and film thickness (0.35–4.06 μm) were proposed. Simulated by a Sudan IV extraction process, the gas–liquid–liquid extraction showed 16–28.3 times working efficiency compared to liquid–liquid extraction, appropriate for application requiring fast analysis and detection.

---

The liquid–liquid extraction process, which is driven by unbalanced chemical potentials between the two liquid phases, is currently used in the field of analytical chemistry. For an accurate analysis of trace reagents in water, liquid–liquid extraction can be used to separate the target reagent with an oil solvent for further detection.¹ Although the enrichment capacity and the target selectivity (determined by the partition coefficient between the phases) are important, the reduction of chemical waste and increased testing efficiency are also critical for the more and more urgent demand of fast screening and detection.² Microchannel extraction^3,4 based on droplet microfluidic processes^5–7 is a fast developing extraction technology. Researchers have found the mass transfer coefficient increases over 100 times in microchannels compared to common liquid–liquid extraction contactors.^8,9 The microdroplets flowing in the microchannels only have nano-liter or pico-liter volumes¹⁰ and the devices themselves are very small, which are favorable for portable and fast detection, one of the main concepts of micro-extraction technology.¹¹ However, the extraction process used in detection is usually operated at extreme phase ratio,¹² which is not in the best operating region from the view point of mass transfer rate. The simple reasons are the long mass transfer distance and the weak circular flow in continuous phase when operated in the Tayler flow with long liquid plugs. Further reducing the droplet size to increase the mass transfer area is not a practical method for its higher break-up demand, such as using particle-packed microchannel¹³ or external field,¹⁴ which increases the detection cost, raises sample filtration demand and adds difficulty in the subsequent phase separation process with much longer droplet coalescence time.

Using hollow droplets is a potential method to solve the problem of extreme phase ratio and enhance the mass transfer rate in extraction process.¹⁵ This capsule droplets, also named multiphase double emulsion,¹⁶ starts reported in recent years and almost all the purpose of this emulsion is preparing core–shell materials,¹⁷ which are of great interest in the areas of cell encapsulation,¹⁸ hollow material¹⁹ and drug delivery.²⁰ Several attempts have been reported to generate the gas–liquid–liquid double emulsion with an inert gas core either using consecutive microstructures, such as flow-focusing, T-junction and co-flowing microchannels,^21,22 or developing unique microstructures, such as double co-flowing microchannel²³ and flow-focusing capillary.²⁴ Among these studies, a special flow pattern is the gas–liquid–liquid double emulsion with thin liquid film on bubbles. This film layer has been successfully used to prepare thin membrane from polymerization²³ or assembling nanoparticles,²⁵ which are potentially useful for encapsulation, delivery of drugs, cosmetics, and nutrients.

In this study, a new microchannel extraction technology named ultra-thin liquid film extraction was developed based on the gas–liquid–liquid double emulsion with a thin middle phase film. The basic principle of this novel method is embedding an inert gas core in the dispersed solvent and making it spread on the bubble surface. The film was controlled to micrometer scaled thickness to reduce mass transfer distance. Since the circular flows in the continuous phase are also enhanced with the shortening of droplet distance, the mass transfer rate can be enhanced number of times.

The gas–liquid–liquid double emulsion was generated in a microfluidic chip shown by Fig. 1. In this chip, a silica tube (D_in-S = 250 μm, D_out-S = 430 μm) was placed into a tapered glass capillary and the tapered capillary (D_in-G = 750 μm, D_out-G = 1000 μm, d_in-G = 100 μm, d_out-G = 160 μm) was embedded in a cross-junction microchannel fabricated on PMMA plate with nearly equivalent width and height (W = 1050 μm, H = 970 μm). In order to control the gas and oil to an ideal non-dispersed flow²⁶ in the trapped capillary, the flow rates was carefully tested. The gas and oil phases ruptured together at the outside of the capillary tip, then the segmented three phases flew into the downstream microchannel. A monitor chamber with rectangle cross-section (W_m = 2000 μm, H_m = 970 μm) was designed to observe the three-phase flow at low velocity. This microfluidic device was similar as our previous double co-flowing microfluidic device,²³ whose inner tube and tapered capillary had aligned ends. The advantage of this improved microchannel was easier to assemble without strict diameter selection of tube and capillary, and the demand for their concentricity was weak too. A Teflon tube was connected to the end of the monitor chamber to provide an extraction time. The total volumes of the microchannel, the monitor chamber and the Teflon tube were 172 μL. Details of device fabrication and structure parameters is provided in the ESI.† A standard microfluidic experimental platform, organized by an optical microscope with high speed COMS camera, two syringe pumps and a low pressure gas feeding system in Fig. 1c, was used to test the new extraction process at room temperature of 24 °C.

Fig. 1 Schematic diagrams of the microfluidic device and the experimental platform. (a) The 3D structure of the device; (b) the flow of three phases; (c) the experimental set-up.

The main components of the working systems used in the microfluidic chip was air, n-octane and water. In order to controllably generate the gas–oil–water styled double emulsion, the interfacial tensions of three phases should satisfy the criterion of γ_GW > γ_OW + γ_GO,²⁵ where corner marks G, O and W refers to gas, oil and water. 0.5 wt% sodium dodecyl sulfate (SDS) were added into water and different concentrations' silicone oil (50 mPa s, x_si = 10–50 wt%) were used in n-octane to adjust the three-phase interfacial tensions and the results are provided in Fig. S1, the ESI.† 1.0 wt% polyvinyl alcohol (PVA, Mn ≈ 83 000) was also added in the water phase to increase its viscosity and stabilize the emulsion. The physical properties of all working systems are listed in Table S1, the ESI.†

For the controllable application of the gas–liquid–liquid double emulsion contained ultra-thin liquid film, it is important to understand the scaling laws of the bubbles and the films. The experiment was first focused on the effects of three-phase flow rates and the composition of oil. Fig. 2a shows a typical three-phase flow at the dispersion position of the microfluidic device. It is clear that the oil flew out of the gap between the inner tube and the outer capillary. In the tapered capillary, the gas phase with black color was almost full of the trapped capillary. Considering the compressibility of gas, the shape of gas–oil interface was not strictly stable. Sometimes, earlier gas break-up in the tapered capillary was observed, especially in the experiment with low gas flow rate. However, no seriously bad effect of this earlier gas break-up was observed in analyzing the bubble size distribution. Interestingly, bubble coalescence was observed at the tip of trapped capillary, rebuilding uniform double emulsions. This was a new phenomenon, which has not been reported according to our knowledge. More details of this rebuilding process are provided in Fig S2, the ESI† and these details show that although a small bubble was earlier generated, it was not big enough to obtain enough force from the continuous phase, therefore the oil film did not break and the bubble was trapped by the film. The trapped bubble then coalesced with the following gas, which was very easy to realize in the present low viscous oils (μ < 4 mPa s) without surfactant. The oil film ruptured until the bubble became big enough, as the force overcame the interfacial tension. This force balance was independent with the earlier gas break-up, thus the double emulsion had almost uniformed sizes.

Fig. 2 Experimental pictures and size laws of bubbles and oil films. The scale bar is 1 mm. (a) Generation of gas–liquid–liquid double emulsion: x_si = 40 wt%, p_G = 0.036 MPa, Q_G = 3620 μL min⁻¹, Q_O = 30 μL min⁻¹, Q_W = 600 μL min⁻¹; (b) in the microchannel: x_si = 20 wt%, p_G = 0.028 MPa, Q_G = 1720 μL min⁻¹, Q_O = 20 μL min⁻¹, Q_W = 600 μL min⁻¹; (c) in the monitor chamber: x_si = 40 wt%, p_G = 0.036 MPa, Q_G = 3620 μL min⁻¹, Q_O = 30 μL min⁻¹, Q_W = 600 μL min⁻¹; (d) on the glass slide: x_si = 30 wt%, p_G = 0.03 MPa, Q_G = 1560 μL min⁻¹, Q_O = 10 μL min⁻¹, Q_W = 600 μL min⁻¹; (e) the dimensionless bubble volume equivalent diameters; (f) the average film thicknesses.

Since the oil films were very thin, it was hard to observe them in the microchannel and the monitor chamber as shown by Fig. 2b and c. Thus, the measured diameters from recorded pictures were deemed as the bubble diameters. Some generated double emulsions were also collected to glass slides and microscope pictures were captured such as Fig. 2d, which exhibited most bubbles had equal diameters. Since the oil film was exposed to atmosphere, the bubble edges looked blurrily due to the volatilization of octane. The volume equivalent diameters of bubbles (d_m) used to characterize their sizes were calculated from the average bubble diameters (d_av) measured from the recorded pictures of the monitor chamber,

where, H_m is the height of monitor chamber. The volume equivalent diameters were ranged from 0.73 to 1.08 mm and they were mainly determined by the flow rate ratio of dispersed phases to continuous phase. Eqn (2) can be used to predict these diameters, as shown by Fig. 2e.

d_m/d_e = 0.068(Q_G + Q_O)/Q_W + 0.6 (2)

where, d_e is the hydraulic diameter of microchannel, d_e = 2WH/(W + H). The size law of eqn (2) is similar as the law of squeezing flow in a T-junction microchannel,²⁷ but it is different from our pervious results for the bubbles generated in a similar co-flowing microfluidic device,²⁸ which contained the effect of capillary number of water phase. This size law and the bubble coalescence phenomenon at the capillary tip shows a different fluid break-up mechanism in the gas–liquid–liquid microflow with high gas flow rate, which is more similar as a spontaneous break-up mechanism but not the vicious shearing ruptured mechanism.

Since it was hard to directly measure the film thickness from microscope pictures, the average film thickness (h_f) was calculated from the operating flow rates and the bubble diameters using eqn (3).

These film thicknesses can be well predicted by eqn (4)

Fig. 2f shows the average film thicknesses ranged from 0.35 to 4.06 μm. The smallest film thickness accords with the result of Kim's minimum 0.62 μm (ref. 29) using a similar dispersion method for the liquid–liquid–liquid systems. However, the uniformity of this film thickness was still waiting to investigate and it was not considered in the present paper for an extraction application. The oil composition had little effect on the bubble size and film thickness laws as shown in Fig. 2e and f. This was because the low phase ratio of oil phase, which did not affect the vicious force every much.

The application of the gas–liquid–liquid double emulsion with ultra-thin liquid film was tested by a 1-phenylazo-2-naphthalenol (Sudan IV) extraction process from water to oil. Since Sudan IV is more soluble in alkane (solubility is 0.216 g/100 g octane) than silicone oil (0.0012 g/100 g silicone oil) at 24 °C, oil phases contained 10 wt% and 20 wt% silicone oil were used in the extraction for a relative high enrichment ratio (the Sudan IV concentration ratio in two phases, M = C_oil/C_water, is about 0.23 for the system contained 10 wt% silicone oil and 0.21 for the system contained 20 wt% silicone oil at 24 °C). 0.015 g L⁻¹ Sudan IV aqueous solution was used as a model system to simulate polluted water, 1 wt% PVA and 0.5 wt% SDS was then added into this model water to generate double emulsion. The aqueous solution was pink, making the process visible. The extraction process was proceeded in the microfluidic chip and the products were collected from the tube outlet. Control group experiments without gas were made to show the mass transfer enhancement effect of this new extraction technology.

All the comparing experiments show obvious mass transfer intensification using the gas–liquid–liquid ultra-thin liquid film extraction technology and Fig. 3 gives a typical example. The left bottle in Fig. 3a is the sample collected from the experiment of Fig. 3b. It is clear that the color of oil layer is orange. Contrarily, the oil layer in the right bottle is colorless, which was collected in a common liquid–liquid extraction process shown by Fig. 3c. The reason for this phenomenon is clearly. The phase ratio of oil to water was 1/16 and the average diameter of the droplets was 968 μm in Fig. 3c. The distance between droplets was 6.92 mm, which was much larger than the droplet size. However, the average thickness of oil film in Fig. 3b was 2.0 μm and the distance between bubbles was zero. Therefore, both mass transfer distances in oil and water phases were significantly decreased and the effect of circular flow in the water phase would be also enhanced with the reduction of liquid plug length. The phase ratio of dispersed phases to continuous phase in Fig. 3b was 5/1.

Fig. 3 Comparison between gas–liquid–liquid double emulsion extraction using ultra-thin film and liquid–liquid extraction using microdroplets. (a) Picture of the extraction products collected at the outlet of the microfluidic chip. The left one is from an ultra-thin liquid film extraction as shown in (b), and the right one is a droplet extraction shown by (c); (b) the working system is x_si = 10 wt%, p_G = 0.029 MPa, Q_G = 1990 μL min⁻¹, Q_O = 25 μL min⁻¹, Q_W = 400 μL min⁻¹; (c) the working system is x_si = 10 wt%, Q_O = 25 μL min⁻¹, Q_W = 400 μL min⁻¹.

Although the oil–water contact time of gas–liquid–liquid system in the microchannel was less than 20% contact time of the liquid–liquid system (residence time was 4.3 s in the experiment of Fig. 3b and 24.3 s in the experiment of Fig. 3c) and the phase separation time became much shorter due to the high density difference of dispersed and continuous phase, the extraction extent of the ultra-thin film extraction process was still much better. Table 1 gives the Sudan IV concentrations in oil measured by an UV-visible spectrophotometer at 513 nm (UV-2450, Shimadzu). This table shows the Sudan IV from the ultra-thin film extraction process is detectable using common UV-visible light detector but the oil absorbency in the control group experiment using droplet extraction is too weak to measure. The enhancement index (EI) in Table 1 is defined as the absorbency ratio of oil phases in different extraction system (A_GLL/A_LL), which indicates the working efficiency can be raised one order of magnitude using the ultra-thin liquid film extraction technology. The distances between bubbles were zero in all the experiments of Table 1 and the film thicknesses exhibit that using thinner film increases the enhancement index. The mass transfer Murphree efficiencies of the gas–liquid–liquid extraction testes defined by eqn (5) were from 62.5% to 80.6%, which were good enough for the application of analysis and detection.

Table 1 Comparison of gas–liquid–liquid and liquid–liquid extraction processes

xsi (wt%)	pG (MPa)	QO (μL min^−1)	QW (μL min^−1)	hfilm (μm)	A	Coil (g L^−1)	EI
10%	—	25	400	—	0.0064	1.17 × 10^−4	18.5
10%	0.029	25	400	2.0	0.1182	2.16 × 10^−3
10%	—	18	320	—	0.0085	1.55 × 10^−4	16.0
10%	0.028	18	320	1.8	0.1363	2.49 × 10^−3
20%	—	30	320	—	0.0044	0.69 × 10^−4	19.1
20%	0.037	30	320	2.1	0.0840	1.97 × 10^−3
20%	—	20	440	—	0.0038	0.81 × 10^−4	28.3
20%	0.030	20	440	1.4	0.1075	2.54 × 10^−3

---

In practical, if the water sample contains oil soluble pollutant, like Sudan IV in this study, and it can be measured by common detectors such as spectrophotometer and chromatography, liquid–liquid extraction becomes a good technology to separate the pollutant from water. The advantage of this gas–liquid–liquid ultra-thin film extraction technology is that it increases the working efficiency, which fits the key point of fast analysis and detection.

In summary, a highly efficient extraction technology based on the gas–liquid–liquid double emulsion contained ultra-thin liquid film was developed. The double emulsion was controllably produced with an improved capillary microfluidic chip, which core structure included an inner silica tube, a tapered glass capillary and a cross-junction microchannel. The average diameters of bubbles generated in this microfluidic device were ranged from 0.73 to 1.08 mm, and they had a linear relation with the phase ratio of dispersed phases to continuous phase. The average film thicknesses were ranged from 0.35 to 4.06 μm. Such gas–liquid–liquid double emulsion with ultra-thin liquid film significantly reduced the mass transfer distances in solvent and water phases, raising the extraction amount of Sudan IV more than ten times comparing to the common liquid–liquid microchannel extraction process. This new extraction technology has strong potential in fast and qualitative screening of trace pollutant in water. Further work for the mass transfer models in this new process will keep on studying in our group.

Acknowledgements

We would like to acknowledge the supports from the National Natural Science Foundation of China (U1302271, 91334201, 21106076) and the National Excellent Doctoral Dissertation Author Foundation of China (FANEDD 201349) on this work.

References

1. N. Assmann, A. Ładosz and P. Rudolf Von Rohr, Chem. Eng. Technol., 2013, 36, 921–936.
2. D. Ciceri, J. M. Perera and G. W. Stevens, J. Chem. Technol. Biotechnol., 2014, 89, 771–786.
3. J. G. Kralj, H. R. Sahoo and K. F. Jensen, Lab Chip, 2007, 7, 256–263.
4. P. Mary, V. Studer and P. Tabeling, Anal. Chem., 2008, 80, 2680–2687.
5. C. N. Baroud, F. Gallaire and R. Dangla, Lab Chip, 2010, 10, 2032–2045.
6. S. Zeng, B. Li, X. O. Su, J. Qin and B. Lin, Lab Chip, 2009, 9, 1340–1343.
7. Q. Zhang, S. Zeng, J. Qin and B. Lin, Electrophoresis, 2009, 30, 3181–3188.
8. Y. C. Zhao, G. W. Chen and Q. Yuan, AIChE J., 2007, 53, 3042–3053.
9. Y. H. Su, G. W. Chen, Y. C. Zhao and Q. Yuan, AIChE J., 2009, 55, 1948–1958.
10. M. Kumemura and T. Korenaga, Anal. Chim. Acta, 2006, 558, 75–79.
11. M. Y. Gao, Y. C. Lu and G. S. Luo, Anal. Chim. Acta, 2009, 648, 123–127.
12. Y. H. Choi, Y. S. Song and D. H. Kim, J. Chromatogr. A, 2010, 1217, 3723–3728.
13. Y. H. Su, Y. C. Zhao, G. W. Chen and Q. A. Yuan, Chem. Eng. Sci., 2010, 65, 3947–3956.
14. S. H. Tan, B. Semin and J. Baret, Lab Chip, 2014, 14, 1099–1106.
15. J. Yue, E. V. Rebrov and J. C. Schouten, Lab Chip, 2014, 14, 1632–1649.
16. N. Deng, W. Wang, X. Ju, R. Xie, D. A. Weitz and L. Chu, Lab Chip, 2013, 13, 4047–4052.
17. W. Wang, C. Yao, M. J. Zhang, X. J. Ju, R. Xie and L. Y. Chu, J. Phys. D: Appl. Phys., 2013, 46, 114007.
18. S. W. Choi, Y. Zhang and Y. N. Xia, Adv. Funct. Mater., 2009, 19, 2943–2949.
19. J. Wan and H. A. Stone, Langmuir, 2012, 28, 37–41.
20. A. R. C. Duarte, B. Unal, J. F. Mano, R. L. Reis and K. F. Jensen, Langmuir, 2014, 30, 12391–12399.
21. J. H. Xu, R. Chen, Y. D. Wang and G. S. Luo, Lab Chip, 2012, 12, 2029–2036.
22. J. D. Wan and H. A. Stone, Soft Matter, 2010, 6, 4677–4680.
23. R. Chen, P. F. Dong, J. H. Xu, Y. D. Wang and G. S. Luo, Lab Chip, 2012, 12, 3858–3860.
24. T. Brugarolas, B. J. Park, M. H. Lee and D. Lee, Adv. Funct. Mater., 2011, 21, 3924–3931.
25. M. H. Lee, V. Prasad and D. Lee, Langmuir, 2010, 26, 2227–2230.
26. B. Y. Xu, S. W. Hu, X. N. Yan, X. H. Xia, J. J. Xu and H. Y. Chen, Lab Chip, 2012, 12, 1281–1288.
27. V. van Steijn, C. R. Kleijn and M. T. Kreutzer, Lab Chip, 2010, 10, 2513–2518.
28. K. Wang, L. S. Xie, Y. C. Lu and G. S. Luo, Chem. Eng. Sci., 2013, 100, 486–495.
29. S. H. Kim, J. W. Kim, J. C. Cho and D. A. Weitz, Lab Chip, 2011, 11, 3162–3166.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra14489a

---