High-Throughput Aqueous Two-Phase System Droplet Generation by Oil-Free Passive Microfluidics.

Aqueous two-phase system (ATPS) droplet generation has significant potential in biological and medical applications because of its excellent biocompatibility. However, the ultralow interfacial tension of ATPS makes droplet generation extremely challenging when compared with the conventional water-in-oil (W/O) system. In this paper, we passively produced ATPS droplets with a wide range of droplet size and high production rate without the involvement of an oil phase and external forces. For the first time, we reported important information of the flow rate and capillary (Ca) number for passive, oil-free ATPS droplet generation. It was found that the range of Ca numbers of the continuous phase under the jetting flow regime is 0.3-1.7, as compared to less than 0.1 in the W/O system, indicating the ultralow interfacial tension in ATPS. In addition, we successfully generated ATPS droplets with a radius as small as 7 μm at the maximum frequency up to 300 Hz, which has not been achieved in previous studies. The size and generation frequency of ATPS droplets can be controlled independently by adjusting the inlet pressures and corresponding flow rates. We found that the droplet size is correlated with the pressure and flow rate ratios with the power-law exponents of 0.8 and 0.2, respectively.

Introduction

The high efficacy of microdroplets for chemical reaction, material synthesis, drug delivery, and disease treatment has been proven in several studies over the last few decades.[1−9] In recent years, the water-in-oil (W/O) system because of its appropriate physical properties has been mainly used to generate aqueous droplets in an oil environment.[1] Despite the remarkable advantages, some important issues need to be considered when using oil as the continuous phase in the manufacturing of aqueous droplets. One of the critical challenges is the recovery of the aqueous sample from the oil phase, which is very difficult and often requires expensive and cumbersome postprocessing.[10,11] Furthermore, the involvement of oil would be detrimental to the viability of cells in biochemical reactions, biomaterial syntheses, and cell or protein encapsulation and always raises concerns about the toxicity of oil.[10−13]

To circumvent these challenges, aqueous two-phase system (ATPS) has been proposed to produce oil-free and nontoxic aqueous droplets.[14] The biocompatible nature of ATPS makes it the best candidate suited for the various biological applications.[13] However, ATPS droplet generation and manipulation are challenging and difficult when compared with the conventional W/O system. This is attributed to the ultralow interfacial tension of ATPS, often less than 0.1 mN/m. It should be noted that the interfacial tension of the conventional W/O system ranges from 1 to 40 mN/m.[15] The low interfacial tension in ATPS results in either a long stream of the dispersed phase throughout the channel without droplet breakup or extremely unstable nonuniform droplets.[10,13,16] As a result, involvement of external forces is always required to facilitate breakup of the stable dispersed phase thread.[16−20] However, these techniques are ineffective in costs and time and involve additional requirements for chip fabrication and experimental setup. In addition, the droplet generation frequency (or throughput) in the active methods, which is of great importance for the scale-up of droplet manufacturing in practical applications, is limited to as high as 50 Hz because of its long response time.[15]

Recently, a passive liquid-filled pipette tip method has been introduced to generate ATPS droplets without the requirement of the external perturbation.[10,13,21] This system has successfully provided continuous passive generation of the ATPS droplets because the interfacial tension force is comparable with the inertial and viscous shear forces. One of the major issues in the previous results is that the inlet flow conditions varied because of the changes in the height of a fluid column in the pipette tip as the fluid was kept injected and the total throughput was limited to the amount of fluid stored in the pipette tip. In addition, no studies of passive ATPS droplet generation reported the exact information of capillary (Ca) number. This is because all previous studies injected fluids using a pressure-controlled mechanism by the pipette tips. Ca number is the most important dimensionless number in microfluidic droplet generation used mainly for studying the fundamental physics, identifying different flow regimes, calculating droplet properties, and most importantly designing the droplet generators with different channel geometries.[1,10,22−25] Because of its significance, many studies with W/O systems provided Ca number and it is typically in the range of 10–3 to 1.[25] Another issue is that the droplet generation frequency is limited at 15 Hz, which is too low to manufacture droplets for actual biomedical applications. To reach a higher frequency of droplet generation, inlet hydrostatic pressures (or flow rates) should be further increased than the current level. Zhou et al.[15] passively obtained high frequencies of ATPS droplet generation by using an additional oil-chopper phase, but their method requires postprocessing for product collection to separate the ATPS droplets from the oil.

In this work, we used precise control of flow rates to produce ATPS droplets without applying any external forces and employing an oil phase. For the first time, we measured inlet flow rates of ATPS during droplet generation, which was impossible in the previous passive oil-free ATPS studies. The flow rate information was used to characterize the droplet properties in terms of the dimensionless Ca number and droplet size. The pressure-/flow rate-controlled passive method generated droplets as small as 7 μm in radius at the generation frequency up to 300 Hz. In addition, we found that the droplet size is correlated with the pressure and flow rate ratios with the power-law exponents of 0.8 and 0.2, respectively. To our knowledge, this is the first successful passive generation of oil-free ATPS droplets with a wide range of sizes at high frequencies, which are comparable with those in the W/O system.

Results and Discussion

Passive Generation of ATPS Droplets by Rayleigh–Plateau Instability

Droplet flow regimes generally consist of squeezing, dripping, jetting, and stratified flow.[1,22] Among them, the jetting flow regime is preferred to obtain spherical droplets with higher throughput. ATPS droplet generation under the jetting flow regime occurs by thread breakup because of the Rayleigh–Plateau (R–P) instability that explains why a falling liquid jet breaks into tiny droplets. The R–P instability growth rate (ω) is a function of the perturbation wavenumber (k), interfacial tension (σ), viscosity (μ), size of the jet (r0), and channel height (h), as shown in eq . The interfacial tension of ATPS is typically in the range of 10–1 to 10–4 mN/m, indicating the ultralow interfacial tension force in droplet generation when compared with that in the W/O system.[13] The ultralow interfacial tension σ would induce a slow growth rate ω along the jet as observed in eq .[15,26] Consequently, it is expected that the droplet breakup length and time are greater and the breakup rate is smaller in ATPS than those in W/O systems. This makes either erratic droplet breakup very far downstream or a thread that survives indefinitely inside the channel without ever breaking up in ATPS.[13,26,27]

Because the interfacial tension of ATPS is much lower than that of the W/O system,  viscous shear and inertial forces should be low to be comparable or less than the interfacial tension force to enable droplet generation. It should be noted that Ca number is the ratio between viscous shear force and interfacial tension force. In other words, the flow velocity of the ATPS should be close to the viscocapillary velocity (Ca number should be less than or about 1; Ca = μv/σ, where μ is the viscosity, v is the flow velocity, and σ is the interfacial tension).[15] Therefore, it is necessary to maintain the ultralow flow rates to break up the dispersed jet passively and generate spherical ATPS droplets. Conventional syringe pump-based passive methods that the majority of W/O systems employed are unable to produce ATPS droplets because the required flow rate is near the lower limit of the flow rate of these commercial pumps. It should be noted that many syringe pumps are unstable at their lower limit of the flow rate.[13]

To overcome this issue, as aforementioned, the pipette tip-based passive droplet generation has been attempted at lower flow rates than that in syringe pumps. One limitation of this method is its inconsistent inlet pressure in given settings because of the change in the height of a fluid column and inherent low frequency because of the fluid injection at too low flow rate. To generate droplets with uniform properties (size and composition), it is necessary to maintain consistent fluid injection at constant flow rates. In addition, to increase generation frequencies, the fluids should be introduced at higher pressure and flow rates than the previous pipette tip ATPS droplet generation.[10,13] In this study, a precise pressure-driven flow injection scheme was utilized to enable generation of rapid, stable, and pulseless droplets with flow rates near the lower limit of commercial syringe pumps but higher flow rates than the pipette tip method. The experimental setup consists of a pressure controller, fluid reservoirs, flow units, and a microfluidic chip (see Experimental Section). Inlet pressure of the dispersed phase (PDEX) ranging from 7 to 15 kPa and inlet pressure of the continuous phase (PPEG) dependent on PDEX varying between 10 and 32 kPa were introduced at constant rates to generate uniform ATPS droplets. It should be noted that the operating pressure in the pipette tip method was close to 1 kPa.[10,13] At fixed PDEX, the maximum and minimum values of PPEG for droplet generation were monitored. For example, at PDEX = 15 kPa, the range of PPEG resulting in droplet generation is 23–32 kPa. PPEG higher than 32 kPa and less than 23 kPa leads to the backflow of dextran (DEX) and the stratified flow of DEX, respectively. At this range of inlet pressure of polyethylene glycol (PEG) and DEX, the resultant flow regime is the jetting in which the generated droplets are smaller with higher frequencies than the dripping regime.

Ca Number and Flow Rate Ratio in ATPS

Information about flow conditions and relevant dimensionless numbers is important to study the fundamental physics and hydrodynamic mechanisms of ATPS droplet generation.[1] In addition, the dimensionless geometry-independent results of the droplet properties are important in application of the present method in various geometries.[10]Ca number, expressed as the viscous shear force over the interfacial tension force between two phases,[28] is the most important parameter in droplet microfluidics by which we can compare the results of this study with different systems of droplet generation such as W/O systems in different configurations.[22] Hydrostatic pressure-driven injection of fluids was used in all of the previous studies of passive, oil-free ATPS droplet generation. The presence of the two-phase flow (i.e., both continuous and dispersed phases) in the main channel of length L makes it impossible to use the relationship between the pressure drop and the flow rate of single-phase flow .[29] This results in no information about the flow rate of both continuous and dispersed phases and relevant Ca number.

Here, we present our test results in terms of the flow rate and Ca number in passive, oil-free ATPS droplet generation (Table ). In conventional W/O systems, because the dispersed and continuous phases were mostly controlled by commercial syringe pumps at constant flow rates, the dimensionless Ca number could be calculated at the typical range of 10–3 to 1. On the other hand, none of passive and oil-free ATPS studies except the current study have provided the Ca number because the inlet flow rates (or velocity) were unknown. In general, the interfacial tension of ATPS is several orders of magnitude smaller than that of W/O systems, and this is why the Ca number in the current study is in the relatively large magnitude when compared with those in W/O systems.

Table 1

Comparison of Ca Number and Flow Rate Ratio for ATPS and W/O System (d: Dispersed, c: Continuous)

		Ca number	flow rate ratio (Qd/Qc)	viscosity ratio (μd/μc)	interfacial tension (mN/m)	aspect ratio (h/W)	refs
ATPS	oil-free passive	0.3–1.7	0.002–0.056	4.2	0.1	0.75	current study
				4.2	0.1	0.5	(10)
				3.8, 4.2	0.037, 0.103	0.33	(13)
	oil-involved passive		0.02–0.13	4	0.1–27		(15)
	active			11.5	0.01	0.33	(16)
			0.1–0.5	2.6	0.1	0.85	(18)
				4.78	0.3		(30)
			0.05–0.2	0.6	0.01	variable	(20)
W/O system (jetting regime)		0.1	0.0025–0.25	0.16		0.12	(31)
		0.1–0.6	0.1–1	1.16	2–5	1	(32)
		0.1	0.025–0.05	0.05	37.76		(33)
		0.002–0.03	0.01–1	265	27	1	(34)
		0.03–0.7	0.05–0.4	0.085–0.18	5–50	1.22	(35)
		0.02–0.1	0.025–4	0.02–16.9	22.1–30.6	1	(36)

As in W/O systems, the relationship between flow conditions and geometries, such as droplet size versus flow rate ratio and Ca number was investigated.[34,35,37−39]Figure a shows variations in the flow rate ratio versus the pressure ratio. As can be seen, the flow rate ratio has a power-law dependence on the pressure ratio in a log–log graph, QDEX/QPEG = 0.3(PDEX/PPEG)3.8. The power and coefficient depend on the hydraulic resistance in the flow path, which is a function of the channel size and viscosity.[4]Figure b shows variations of dimensionless droplet size (droplet diameter, D) versus the flow rate ratio. As is clear from the figure, increasing the flow rate ratio results in an increase in the droplet size. This is in line with the fact that increasing QDEX and decreasing QPEG leads to bigger droplet generation. D/W varies between 0.15 and 0.28 when the flow rate ratio linearly increases from approximately 0.002 to 0.056. This linear relationship has also been observed by Garstecki et al.[40]

Figure 1

(a) Variations of the flow rate ratio as a function of the pressure ratio and (b) dimensionless droplet size vs the flow rate ratio.

Figure presents the droplet size and CaDEX in terms of CaPEG. Figure a provides unique contributions of viscous shear and interfacial tension forces to droplet generation as the Ca number is directly related to these two forces. It should be noted that the Ca number provides a unique venue to determine the droplet properties in the various systems. The range of CaPEG in our experiments is 0.3–1.7 (Ca ≈ O(1)), indicating that the ultralow interfacial tension of the system and the Ca number in the current study are greater than those in the conventional W/O systems (Table ). This implies that the effect of the viscous shear force exerted by the PEG phase is comparable to the interfacial tension force. We observed that the dripping flow regime in the ATPS droplet generation does not occur in this range of Ca numbers, but rather, this range of Ca numbers is favorable for the jetting flow regime. The unique ultralow interfacial tension in ATPS makes it impossible to generate droplets close to the junction at even low flow velocity as CaPEG is still around 1, which should be in the jetting regime (Table ). In such conditions with extremely low velocity, that is, low inlet pressure and low inlet flow rate of PEG, bigger droplets at low frequencies are generated under the dripping flow regime, which is not a favorable condition for droplet generation.[10] On the other hand, to achieve the high frequencies of droplet generation, an increase in CaPEG number at fixed PDEX makes droplet size decrease because of the increase in the velocity of the PEG phase and resultant enhanced viscous shear force, as shown in Figure b. From these figures, it is observed that the minimum value of Ca number in the jetting flow regime is about 0.2, as found in the literature,[41] whereas the maximum Ca number is close to 1.7. It should be emphasized that the range of Ca numbers in the jetting flow regime in ATPS is introduced in the current study (Table ).

Figure 2

(a) Variations of CaDEX vs CaPEG and (b) droplet size vs CaPEG.

Droplet Generation Frequency and Droplet Size

Higher throughput with superior size controllability, leading to higher manufacturing efficiency and significant savings of chemicals and time, is an important need for many large-scale applications. Table shows a comparison of the droplet size and resultant generation frequency in various ATPS droplet generation schemes. In terms of the droplet size, the current study provides much smaller droplets when compared with those in active methods while compatible with passive methods. The droplet size was conveniently controlled by changing the inlet flow rates of two phases. On the other hand, the generation frequency was significantly improved compared with the previous reports in ATPS droplet generation.

Table 2

Droplet Size and Generation Frequency in the Previous Research Studies of ATPS[15]

method		droplet radius (μm)	generation frequency (Hz)
passive method	current study	7–14	∼300
	pipette tip[13]	∼5 to ∼55	∼15
	pipette tip[10]	∼20 to ∼70	<3
active method	piezoelectric bending disc[18,30]	12.5–37.5	<50
	pin actuation[20]	∼60 to ∼93	∼2.5
	mechanical vibration[17,45]	20–100	∼30
	pulsating inlet pressure[16]	22–177	∼5
	electrohydrodynamic perturbation[19,46]		∼5

Figure shows the droplet frequency with respect to PPEG at different PDEX. In this figure, the droplet generation frequency increases with increasing PPEG at all PDEX. The highest droplet frequency is found at ∼300 Hz at PDEX = 11 kPa and PPEG = 30 kPa, which has not been achieved in all previous ATPS droplet generation (Table ) (see Supporting Information Video S2). High generation frequency of ATPS droplet generation is essential for laboratory or clinical settings such as biological and chemical assays in droplets and high throughput manipulation, analysis, sorting, and encapsulation of cells.[42−44] To increase the droplet generation frequency further, PPEG should be increased in theory. However, considering the dimensions of the microchannel in our experimental setup, further increasing of PPEG results in a very thin stream of DEX and unstable tiny droplets.

Figure 3

Variations of the droplet frequency as a function of PPEG at different PDEX.

Small aqueous droplets in picoliter to nanoliter are particularly useful for cell research, drug delivery, organic particle synthesis, and microreactors.[47] In droplet microfluidics, it has been known that the drop size typically depends on the channel geometry and input pressures.[10] The channel geometry and input pressures directly alter the dispersed and continuous phase flow rates, making peculiar flow regimes with various droplet sizes. In our study, spherical droplets in the range of 7–14 μm in radius were generated mainly by the R–P instability in the jetting flow regime (see Supporting Information Videos S1 and S2).

Figure a shows the variations of droplet size at different PPEG and PDEX. It shows that the droplet size increases by increasing PDEX at fixed PPEG. This result is well aligned with our observation that the breakup point of the liquid jet moves from the junction to downstream, causing a long DEX stream. In other words, an increase in PDEX causes a higher inertial force of the dispersed phase than the interfacial tension force resulting in larger droplets. Because of the low interfacial tension force, droplets are unable to maintain their perfect spherical shape.[26]

Figure 4

(a) Variations of droplet size as a function of PPEG at different PDEX and (b) dimensionless droplet size vs the pressure ratio.

On the other hand, as PPEG increases at fixed PDEX, the location of droplet pinching-off moves closer to the junction resulting in smaller droplets. The viscous shear force exerted on the interface would be increased by the higher flow rate of the continuous phase, which makes a stronger elongation effect acting on the DEX thread-thinning process and the breakup of the DEX phase into smaller droplets.[48] The smallest droplet size in our study is 14 μm in diameter at PPEG = 30 kPa and PDEX = 11 kPa (Figure a). Figure b shows the dimensionless droplet size versus the inlet pressure ratio. The size of formed droplets can be scaled with the pressure ratio (PDEX/PPEG = 0.37–0.71) of both phases as a power-law relationship (D/W = 0.4(PDEX/PPEG)0.8, Figure b inset). The droplet size as a function of flow rate ratio shows fewer variations than the pressure ratio (in log–log coordinates). From Figures b and 1b, it can be seen that the increase in droplet size is greater by increasing the pressure ratio compared with the flow rate ratio.[37]

Conclusions

Understanding the fundamental mechanisms of droplet generation in all aqueous phases has important implication, particularly in biomedical research. In this paper, we investigated flow characteristics and resultant droplet properties during droplet formation in a flow-focusing geometry in a passive ATPS system without the involvement of an oil phase and external forces. In contrast to all previous passive ATPS studies, we provided Ca number in the range of 0.3–1.7 based on flow rates maintained constant. This range of Ca number in ATPS is more than 3 orders of magnitude larger than the conventional oil–water systems, indicating that ATPS generates droplets at the ultralow value of interfacial tension. The relatively high range of Ca number requires unique inlet flow conditions where flow rates should be set in near the lower limit of commercial syringe pumps and the upper limit of the pipette tip method. The exclusive flow regimes generate ATPS droplets with different sizes, 7–14 μm in radius, and frequencies up to 300 Hz, impossible previously in both passive and active methods for ATPS droplet formation. The size and frequency were significantly improved when compared with those in the previous studies. Particularly, generation of ATPS droplets in the passive method without the involvement of an oil phase has been limited to 15 Hz. The experimental results showed that the droplet size can be characterized by a power law of both pressure and flow rate ratios with different exponents of 0.8 and 0.2, respectively.

Experimental Section

Experimental Setup

A schematic of the experimental setup including a flow-focusing microfluidic chip and a flow control system is shown in Figure . The flow-focusing geometry consists of two inlet and one outlet ports. The width (W) of the downstream and PEG channels is 100 μm, whereas the width of the DEX channel is 50 μm. The height (h) of all of these channels is 75 μm. A high-precision microfluidic pressure control system, MFCS-EZ (Fluigent, Inc., USA), with a flow unit platform including a flow-board and two in-line flow units was used. The flow units are two high-precision individual flow sensors for measuring the flow rate: M model (up to 80 μL/min with a lowest detectable flow rate increment of 0.06 μL/min) and S model (up to 7 μL/min with a lowest detectable flow rate increment of 0.01 μL/min) were used for the PEG and DEX phases, respectively. As shown in Figure , the outlet of the pressure controller was connected to fluid reservoirs of PEG and DEX to inject controlled pressurized air. The pressurized air drives the fluid and then the fluid flows through the outlet of the reservoirs. After passing through flow units, the fluid enters the chip creating droplets. Multiple calibration processes were performed for accurate flow rate measurements. An inverted microscope (IX73, Olympus Corp., Japan) with a 10× objective lens and high-speed camera (INFINITY3-3UR, Lumenera Corp., Canada) was used to capture images of ATPS droplets. For high-speed imaging, the camera operated at 50–80 fps with an exposure time of 1.0 ms. ImageJ software was used for image processing such as droplet size measurements.

Figure 5

Experimental setup and schematic of a microfluidic chip.

Microfluidic Chip

The microfluidic chip was fabricated using standard soft lithography and photolithography techniques with PDMS (polydimethylsiloxane). The channel geometry was created using computer-aided design (CAD) software (AutoCAD 2016, Autodesk, Inc., USA). To form a patterned photomask, the CAD designs were printed onto a transparency sheet (25 400 dpi, CAD/ART Services Inc., USA). Spin coating was used to distribute a thin layer of KMPR 1025 photoresist (MicroChem, USA) to a 4 in. silicon wafer (UniversityWafer, Inc., USA). After exposure to UV light through the patterned photomask, the wafer was chemically developed to form the channel geometries. A 10:1 ratio mixture of PDMS (Sylgard 184, Dow Corning, USA) resin to curing agent was poured onto the silicon wafer. The wafer was allowed to cure in an oven for 1 h before removal of the PDMS layer. Inlet and outlet holes were made using a 1.0 mm diameter biopsy punch (Integra Miltex, Inc., Germany). Finally, the PDMS layer and a glass microscope slide (25 × 75 × 1.0 mm, Fisher Scientific, USA) were permanently bonded through an oxygen plasma treatment (Harrick Plasma, USA).

Chemicals

To prepare the ATPS solution, two solutions of 10% (w/v) PEG (Sigma-Aldrich, USA) and 16% (w/v) DEX (Alfa Aesar, USA) were formulated by dissolving PEG and DEX separately in deionized water. The two solutions were then thoroughly mixed in a beaker using a magnetic stirrer (Isotemp stirring hotplate, Fisher Scientific). The solution was left for a minimum of 24 h of inactivity while phase-separation occurred, forming upper PEG-rich and lower DEX-rich phases. After the solution reached equilibrium, we partitioned the two phases with syringes into separate 50 mL conical centrifuge tubes (Corning Falcon centrifuge tubes, Fisher Scientific). Viscosity values are μPEG = 17.0 mPa·s and μDEX = 71.9 mPa·s for the PEG and DEX phases, respectively.[13] The interfacial tension of the ATPS solution is 0.103 ± 0.006 mN/m, and the densities of the PEG and DEX phases are ρPEG = 1013 kg/m3 and ρDEX = 1067 kg/m3, respectively.[49]