Introduction

Droplet-based microfluidics involves the manipulation of immiscible fluids, such as oil and water, within microchannels to create discrete droplets or slugs. This technique relies on precise fluid injection and flow control to generate and manipulate the droplets1,2,3. By aiding in the formation of stable emulsion, droplet microfluidics can be utilized in a wide range of applications, including biological, pharmaceuticals4, cosmetics5, food and agriculture, and microfluidic sensors6,7,8,9. Depending on the application, droplets composed of water or oil are used. The versatility of droplet microfluidics is attributed to the ability to control droplet size, composition, and flow dynamics with high precision, making it an essential tool in both research and industrial applications. To generate the uniform droplet and enhance encapsulation efficiency, various methodologies have been reported such as magnetic strategy10, osmotic modulation11, compartmentalized synthesis12, microfluidics13.

In droplet microfluidics, surfactants are commonly used to stabilize emulsions by reducing the interfacial tension between immiscible fluids, such as water and oil. Surfactants play a crucial role in preventing droplet coalescence, increasing the viscosity of the continuous phase, matching the density with each other, and improving the overall stability of emulsions14,15,16,17. However, the use of surfactants involves certain risks. Many surfactants, particularly synthetic ones, have raised concerns regarding their potential toxicity when absorbed into the human body or ingested. These compounds can accumulate in living organisms, posing risks to health and the environment18,19,20. Additionally, surfactants may introduce unwanted chemical interactions with biological systems, limiting their safe use in sensitive applications like drug delivery and food products. Many surfaces are not biodegradable and can have harmful effects on ecosystems. As a result, there has been increasing interest in finding alternatives that can stabilize emulsions without relying on surfactants21,22,23,24.

Given the drawbacks associated with surfactants, research has focused on developing methods to form stable emulsions without them. Generally, without surfactants, droplets are unstable and merge with each other. However, as the size of the droplets decreases below a certain threshold, Brownian motion becomes more active than particle movement driven by gravity force, preventing the sedimentation or creaming of the emulsion and thus making the droplets kinetically stable23,25. As a result, many studies have been conducted aiming to reduce droplet size as much as possible by utilizing various principles. One promising approach is ultrasound emulsification, a non-contact method that utilizes high-energy acoustic waves to generate droplets through cavitation21. This process involves two stages21,26: First, the interface between the oil and water phases becomes unstable, leading to the dispersion of the phases. Second, cavitation forces further break up the generated droplets, leading to the formation of smaller ones. In this way, the oil droplet can be divided into smaller droplets. Ultrasound emulsification can efficiently produce nano-sized droplets. However, ultrasound emulsification typically requires high power and multiple rounds of treatment to achieve uniform droplet size and high stability ((i) in Fig. 1).

Fig. 1
Fig. 1
Full size image

The ultrasonic emulsification system composed of (i) recirculating main tube including oil droplets in water solvent, (ii) oil–water segmented flow from the pre-processing microfluidic device, and an ultrasonic emulsification equipment.

To address these challenges, microfluidic techniques can be employed as a pre-processing step, where the oil phase is fragmented into smaller droplets using controlled microfluidic flows. In this study, microfluidic device was designed and utilized for generating oil droplets without surfactants as the pre-stage of emulsification ((ii) in Fig. 1). By achieving a more uniform droplet size in the initial stage, the subsequent ultrasound treatment can be more effective and require lower energy inputs, resulting in a more efficient and controlled emulsification process. This combination of microfluidics and ultrasound offers a promising solution for surfactant-free emulsions with enhanced stability and precision.

Materials and methods

Theory of critical diameter for oil droplet stability

The velocity of the particle due to gravitational force is given by Stokes’ law as following.

$$\text{v}= \frac{({\uprho }_{\text{p}}-{\uprho }_{\text{f}})}{18\upmu }{\text{gd}}^{2}$$
(1)

where \(g\) is the gravitational constant, \(d\) is the diameter of the particle, and µ is the dynamic viscosity. \({\rho }_{p}\) and \({\rho }_{f}\) are the density of the particle and the fluid, respectively. On the other hand, Brownian motion is associated with the thermal energy of the system, which is the random motion of particles suspended in a medium. The mean squared displacement, \(d\) due to Brownian motion in three-dimension is as following27,

$$\text{d }= \sqrt{6\text{Dt}}$$
(2)

where \(\text{ D}\) is the diffusion coefficient of the particle, and \(\text{ t }\) is the elapsed time. The interplay between these two distinct modes of particle movement allows for the definition of a critical particle size. This critical size represents a threshold where the effects of gravitational forces and Brownian motion are comparable. Below this critical size, the randomizing effect of Brownian motion significantly outweighs the directional pull of gravity. As a result, these smaller particles are kept in suspension by their continuous, erratic thermal movements, which effectively counteract sedimentation or creaming. This dominance of Brownian motion for sub-micron particles, such as those in nanoemulsions, is crucial for their long-term stability against gravitational separation and aggregation, allowing them to remain uniformly dispersed in the medium.

Microfluidic device design

A typical microfluidic device for droplet generation uses a T-junction, which induces the dispersed fluid to be destabilized by the momentum change due to shear force of the continuous fluid28. It is related with the interaction between channel dimensions, fluid viscosity, flow rate and pressure29. In this method, when an immiscible fluid in the vertical direction is supplied into a main channel where the carrier fluid is introduced, the fluids separate, and droplets are formed. Figure 2 represents the types of channels based on T-junction for creating segmented flow of oil and water without surfactant. Figure 2a is typical T-junction channel30,31,32. It allows the oil flow to be broken by water flow. To minimize oil contact with the channel surface, a blunt needle in Fig. 2b can be inserted into the center of the channel. The oil through the needle is squeezed by water flow in all radial directions33,34.

Fig. 2
Fig. 2
Full size image

The types of channels for creating segmented flow of oil and water without surfactant. (a) Conventional T-junction channel, (b) Needle-inserted channel, and (c) Needle-inserted glass capillary channel.

By selecting a microchannel material with higher surface energy, such as glass, or by coating the channel with hydrophilic materials, this approach enhances water wettability, reducing interfacial instability and improving the overall efficiency of the separation process. As shown in Fig. 2c, a glass capillary is connected to the T-junction, and oil droplets, separated from water at the needle tip, can flow inside the capillary. The channel fabricated using the PolyJet method is made of VeroClear, a material with surface energy (41.7 mN/m) and contact angles (67.0°)35. On the other hand, the used glass capillary exhibits a water contact angle of around 27°, indicating high hydrophilicity36. The contact angle was measured by using a goniometer based on the sessile drop technique, which is a method of optically analyzing the contact angle between a droplet of liquid and a solid surface.

Fabrication of microfluidic device

The microfluidic device was designed using SolidWorks software (Dassault Systèmes SolidWorks Corp., USA), fabricated by a 3D printer (J850, Stratasys, USA), and then assembled with a blunt needle (27G) and a PTFE (Polytetrafluoroethylene) tube. PolyJet method was adopted as a 3D printing technique, which uses ultraviolet light for curing the liquid photopolymer layers immediately after deposition37. The single printed layer thickness is 14 μm, and the average roughness of the printed surface is less than 10 μm. For the laminar flow condition, the effect of surface roughness on the flow is reduced since the boundary layer is thicker than absolute roughness, and viscous sublayer adheres to and flows along the surface38. In this study, with a Reynolds number (Re) of approximately 400 or less, the printed channel wall can be considered a hydrodynamically smooth surface39.

The needle and glass capillary were inserted into the center of the microfluidic channels and securely fixed using epoxy resin (NOA 68, Edmund Optics, US). For the devices made of other polymer materials such as polypropylene, polyvinyl chloride, polycarbonate, and acrylic, precision machining was used to fabricate the device. The internal diameter of T-channel was 500 μm. The inlet and outlet fluid connections of the device were created by machining UNF 1/4″ 28G thread, to which flangeless male nut with 1/16″ OD tubing were then connected.

Experiments

Experimental setup for generating the oil–water segmented flow is shown in Fig. 3. It is composed of a stereo microscope (Stemi 508, ZEISS, Germany), a high-speed camera (VEO-E 310L, Vision Research, Inc., US), a syringe pump (Fusion 4000, CHEMYX, USA), and a microfluidic chip (Fig. 3a). The stereo microscope with LED light source is connected with a high-speed camera, which captures the flow image through the objective lens. Grape seed oil and DI water were injected into their respective inlets, where they came into contact and broke into discrete segments, resulting in a segmented flow. As shown in Fig. 3b, the dispersed fluid forms droplets which are fragmented by the flow of continuous fluid, while the continuous fluid forms slugs. This process is governed by the interaction between several forces including surface tension, shear stress, and pressure drop40,41,42,43,44,45. Between the two types of fluids, the classification into continuous fluid (slug) and dispersed fluid (droplet) is generally determined by the affinity between the channel surface and each fluid. This affinity dictates which fluid wets the channel surface more effectively, establishing its role as the continuous phase, while the other fluid forms dispersed droplets.

Fig. 3
Fig. 3
Full size image

Experimental setup for investigating the oil–water segmented flow: (a) Experimental setup composed of a stereo microscope, a high-speed camera, a syringe pump, and a microfluidic channel. (b) The concept of generating oil–water segmented flow in the needle-inserted channel, and the illustration of slug (carrier fluid) and droplet (dispersed fluid) of the flow.

Image processing

To analyze the oil–water segmented flow, post-processing of the raw image was conducted using MATLAB (Mathworks, US). The background image was obtained by averaging about 300–6500 images, depending on the frame rate of camera, and then subtracted from the raw image to remove any noises that could affect the intensity data (Fig. 4a). The grayscale image was converted into a binary format to distinguish the interface between oil and water, based on intensity changes. The global binarization threshold was automatically determined using Otsu’s method, which minimizes intra-class variance. The subsequent binarization was performed with this determined threshold. Figure 4b describes the method to analyze the length of segmented fluid. Following image binarization, the precise locations of the interfaces were determined. For each image frame, a signal profile was extracted along the central axis of the microchannel from the binarized image. Abrupt changes in this signal were then detected and adopted as the definitive locations of the oil–water interfaces, marking the beginning and end of each phase segment46. The oil region was assigned a value of 1, and the water was 0. The length of segmented oil was then calculated by dividing the sum of the signal values by the number of oil slugs for an instantaneous flow image. Finally, the average oil length was determined for all captured images, excluding data exceeding 10% of the initial averaged value. To estimate the droplet volume, data processing was performed as in Fig. 4c. The droplet edge was detected from the converted binary image using ImageJ. The image was then subtracted and reloaded into MATLAB. A droplet cross-section was obtained through curve fitting. Assuming that the cross-section with known XY coordinates is axisymmetric, the volume of the droplet was calculated by performing a rotational integral about the x-axis.

Fig. 4
Fig. 4
Full size image

Image processing for droplet analysis: (a) Flow chart of post-processing method. (b) Process of analyzing the length of segmented fluid using the binarized flow images. (c) Process of calculating the volume of droplet.

Pre-processing device for ultrasonic emulsification

In general, larger oil particles require more energy and time for fragmentation, leading to a decrease in emulsification efficiency. By utilizing a carefully designed microchannel, oil droplets are pre-fragmented into smaller sizes before entering the emulsification equipment. This pre-processing step ensures that the initial oil droplets are more uniformly distributed and smaller in size, which in turn reduces the workload on the emulsification system. To quantify the effectiveness of the microfluidic channel as a pre-processing device for generating the oil–water segmented flow, ultrasonic emulsification equipment (DEBREX, FUST Lab, South Korea) was used. As shown in Fig. 1, the oil–water segmented flow (ii) was injected into the recirculating tube (i) and delivered into the emulsified region inside the equipment, where it was broken into smaller droplets due to the cavitation effect. When the pressure exceeded a critical threshold, the oil droplets collapsed, driven by localized shear forces and temperature elevation47. The oil–water mixture was circulated through the ultrasonic region by a peristaltic pump at a flow rate of 20 mL/min for up to 10 min. Oil droplet size after emulsification was measured by the laser diffraction particle size analyzer (Partica LA-960, HORIBA, Japan) after ultrasonic emulsification.

Samples for analysis were obtained at 3, 5, and 10 min after the oil–water segmented flow entered into the ultrasonic equipment. Particle size distributions was compared between the control and pre-processing groups as a function of time.

Results and discussion

Conventional T-junction channel with different materials

Figure 5 shows the generated oil–water segmented flow in the conventional T-junction channel (Fig. 2a). The image was taken by the high-speed camera with 2000 frame per second. Oil enters the main channel, gradually grows in size, and blocks the channel. As the water flows through the main channel, the oil thread is stretched and distorted downstream and squeezed by the water flow. Finally, the oil thread collapses when the surface tension force at the oil–water interface becomes lower than the external forces, such as pressure drop and shear force45.

Fig. 5
Fig. 5
Full size image

Unstable oil–water segmented flow in the conventional T-junctions with different channel materials. For the cases of polypropylene and polycarbonate materials, the oil slugs to be coalescence were marked as (1) and (2).

To investigate the interaction effects of channel materials, four different materials (polypropylene, polyvinyl chloride, polycarbonate, and acryl) were examined. In all cases, oil was generated in the form of slugs rather than droplets, and the flow appeared highly unstable. This instability arises because the oil exhibits a strong affinity with the channel materials, forming a continuous fluid and breaking into slugs48,49,50,51,52. In channels fabricated from polypropylene and polycarbonate, the flow behavior was examined using sequential images captured at different time intervals. These observations revealed that oil slugs tended to coalesce as they moved through the channel. The viscosity of oil is much higher than that of water, and it preferentially wetted the channel surface, leading to increased flow resistance and notably slower velocity relative to the water phase. Consequently, a leading oil slug (marked (1) in the images) was overtaken by and merged with an immediately trailing slug (marked (2)).

Needle-inserted channel

The needle-inserted channel (Fig. 1b) was designed to eliminate the influence of channel material on the oil fragmentation process, enabling the generation of oil in droplet form rather than the slug. The oil flow rate was fixed at 0.3 mL/min, while the water flow rate (Qw) was increased from 0.1 to 1.0 mL/min. Figure 6a shows images of oil–water segmented flow with Qw. When the Qw was 1 mL/min, a segmented flow was formed unstably. At Qw in the range of 0.1–0.7 (mL/min), flow was stable but oil remained in slug form. Figure 6b represents a plot of water and oil size depending on the water flow rate with the standard deviation. As the water flow rate increased, the water droplet size increased whereas the oil size decreased. A significant reduction in oil size was observed when the Qw increases from 0.1 to 0.3 mL/min.

Fig. 6
Fig. 6
Full size image

Oil–water segmented flow at the needle-inserted channel: (a) Flow images of oil–water segmented flow depending on the water flow rate (Qw = 0.1–1.0 (mL/min)). (b) A plot of water and oil size at Qw = 0.1–0.7 (mL/min).

Needle-inserted glass capillary channel

To generate oil droplets rather than slugs without surfactants, a glass capillary was incorporated into the previously used needle-inserted T-channel design since the formation of droplet is affected by the viscous force and the interfacial tension32,53. The Capillary number (Ca = μU/γ) for the flow conditions in the needle-glass capillary was estimated to be within the range of 10–2–10–3, using the interfacial tension between oil and water (25.15 mN/m)54,55. These low Ca values indicate that interfacial tension forces are dominant over viscous shear forces in governing the fluid behavior near the wall and droplet formation. The glass capillary surface is highly hydrophilic and the dominance of interfacial tension effects in this low Ca regime ensures that water preferentially wets the glass surface. This energetically favorable interaction promotes the formation and maintenance of a stable water film on the capillary walls, effectively shielding the oil phase from direct contact with the glass wall and thereby inhibiting oil wetting.

The needle was positioned at the cross-sectional center of the glass capillary, allowing oil and water to interact within the glass channel, rather than contacting the surface of the 3D-printed chip. Water could act as the carrier fluid, while oil would form the dispersed fluid, leading the formation of oil droplets. Figure 7a shows images of oil–water flow in the glass channel depending on Qw. Under the 1 mL/min condition, which previously showed unstable flow in the earlier needle-inserted model shown in Fig. 2b, the combined design demonstrated stable flow with the formation of oil droplets. Moreover, even at a higher Qw of 3 mL/min, a stable flow was observed, and the droplet size further decreased. Figure 7b shows a generated droplet volume as obtained based on the process in Fig. 4c. Each volume of droplet ​​obtained for the 6 cases is 0.72 and 0.44, respectively, at Qw = 1, 3 mL/min, and their sample standard deviations are 0.02 and 0.03, which have a relative standard deviation of less than 7%. At higher flow rates, it was observed that the volume of oil droplets decreased.

Fig. 7
Fig. 7
Full size image

Oil–water segmented flow at the needle-inserted glass capillary: (a) Raw images of oil–water flow in the glass channel depending on the water flow rate (Qw). (b) A plot of oil droplet size, and (c) Preliminary tests for visualization of created oil droplet flow in the emulsifying section connected with a 3D-printed chip and a peristaltic pump.

Ultrasonic emulsification using pre-processing microfluidic device

After verifying that oil droplets were stably formed in the glass capillary model, preliminary tests were performed to evaluate its application in actual ultrasonic emulsification. At the left side of Fig. 7c, an oil droplets flows out of the pre-processing microfluidic chip, while a peristaltic pump injects water into the emulsification section23. To visualize the droplet exit at the end of the tube in the sonification region, a transparent glass tube was installed to replace the ultrasonic transducer. On the right side of Fig. 7c, the generated oil droplets stably flow passing through the glass capillary (1), and are subsequently delivered to the ultrasonic emulsification region (2).

The emulsification results using the needle-inserted glass capillary channel were examined by comparing the oil droplet distribution obtained with and without the use of pre-processing devices. Figure 8 shows the particle size distributions after emulsification. For the pre-processing group, quantities in each fraction were plotted with operation time in Fig. 8a. Samples were measured after 3, 5, and 10 min following the injection of the fluid into the equipment. The most frequently occurring particle size within the distribution, known as ‘mode’, decreases from 4.18 to 2.80 μm with operation time, and the kurtosis of the distribution also increases gradually over time. This indicates that as the operating time increases, the size of the emulsified particles decreases, and the variation in droplet size becomes smaller. Figure 8b,c represent the particle size distributions for the control and pre-processing groups of samples collected at 3 and 10 min of operating time, respectively. In the particle size distribution for the sample of the 3-min operation time (Fig. 8b), the pre-processing group shows a higher proportion of particles around the mode size compared to the control. In contrast, the control group exhibits a relatively asymmetric distribution. While the mode value seems to be same around 4.18 μm, the mean values were 5.85 μm for pre-processing and 7.22 μm for the control group, indicating that the particle deviation in the control group was larger. It represents that the pre-processing group has a more uniform distribution of target particles compared to the control case. In the distribution after the 10-min operation (Fig. 8c), as in the previous results, the pre-processing group exhibits a more uniform distribution for the specific particle size. Both the mean and mode values were smaller in the pre-processing group, measuring 3.04 μm and 2.80 μm, respectively, compared to 3.90 μm and 3.63 μm in the control group, confirming that smaller particle sizes were achieved with a needle-inserted glass device. According to the results, the particle size reduction rate (mode value) was 2.5 times greater in the case of pre-treatment than in the case of control group. As a result, means the time for approaching the target size could be reduced in the case of the presence of pre-processing microchip, as compare with the control case. For that, the power of the equipment used in this experiment is 100 W, and for a pre-processing group, approximately 36 kJ of energy could be saved to create particles of the same size compared to when the pre-processing device is not applied. In addition, the deviation of particle size was enhanced as the kurtosis for pre-processing case was higher than for the control case. In the relevant industrial applications, to create the desired target size, the sample is injected into the emulsification equipment and circulated multiple times over a set period to form the nanoemulsion. A microdevice that generates and injects oil droplets without surfactants can be used in systems aimed at improving the manufacturing efficiency of the ultrasonic emulsification process.

Fig. 8
Fig. 8
Full size image

Particle size distributions after emulsification: (a) Particle size distributions with quantities in each fraction considering operation time for the pre-processing group. Particle size distributions for the sample of (b) the 3-min operation time, and (c) the 10-min operation time.

Discussion

This study demonstrates the effectiveness of using a microfluidic device as a pre-treatment step before ultrasound emulsification, providing a surfactant-free approach to generating uniform oil droplets. First, a conventional T-junction channel was used to create the oil and water segmented flow depending on the various materials such as polypropylene, polyvinyl chloride, polycarbonate, and acrylic materials. Although acrylic material showed relatively better performance compared to others due to higher surface energy influencing the wettability of the channel surface, the form of oil slugs in this model was not suitable for use in ultrasonic emulsification.

The microfluidic chip, designed with a needle positioned at the center of the glass capillary, successfully facilitated the formation of stable droplets without the use of any surfactants. This model could eliminate material-dependent issues, leading to improved formation of oil droplets rather than slugs. The placement of the needle in the channel forces oil droplets to be squeezed in all radial directions by the water flow while the hydrophilic glass capillary prevents coalescence and stabilizes the oil and water segmented flow. By ensuring a consistent droplet size before the ultrasonic emulsification process, both mean and mode values of particle distribution in the pre-processing group are lower than that in the control group. Therefore, the pre-processing microfluidic device leads to smaller droplets and a more uniform size distribution with increased kurtosis after the ultrasonic emulsification. It indicates that this method is expected to improve emulsification efficiency while reducing power consumption.

Despite its advantages, the effectiveness of the microfluidic chip may depend on the specific fluid properties, such as viscosity and interfacial tension, which could limit its applicability to a wider range of emulsions. For instance, the current study utilized grape seed oil, which has a relatively moderate viscosity. If a significantly higher viscosity dispersed phase, such as certain high-viscosity silicone oils, were to be used, the droplet formation dynamics at the needle tip could be substantially different. Higher viscous forces within the dispersed phase would resist breakup, potentially requiring higher continuous phase flow rates to achieve sufficient shear, or modifications to the needle/capillary geometry (e.g., a narrower constriction) to facilitate droplet detachment. Furthermore, the polarity of the dispersed phase could significantly influence the system’s performance. The oil used in the study is non-polar. If a more polar dispersed phase, such as certain alcohols or other polar organic solvents, were employed in an oil-in-water configuration, the interfacial tension between the dispersed phase and water would likely change. A lower interfacial tension might facilitate easier droplet breakup, potentially leading to smaller droplets at similar flow conditions. The wetting characteristics could also be affected. The effectiveness of the water film in preventing dispersed phase adhesion would depend on these competitive wetting effects. If the polarity difference between the two phases is small, or if the dispersed phase has some affinity for glass, the stability of the segmented flow might be compromised.

Future research for advancing this surfactant-free emulsification approach, could be directed towards multi-stage microfluidic emulsification systems, where droplet size is progressively refined through sequential modules, potentially further optimizing the sonification system. To address scalability and enhance throughput for potential industrial applications, the development of an adjustable needle array design for parallelized droplet generation also needs to be investigated. Such systems could offer not only higher production rates but also dynamic control over droplet size distributions by tuning array configurations and individual needle parameters. Furthermore, the integration of particle size monitoring techniques within the microfluidic setup or downstream processing line would enable real-time feedback for precise process control, quality assurance, and facilitate the development of automated emulsification systems56.

The findings of this study have broad implications for industries such as pharmaceuticals, cosmetics, and food processing, where surfactant-free emulsions are highly desirable. The microfluidic-based pre-processing method can be further developed for applications requiring precise droplet control, such as drug delivery systems and encapsulation technologies. From another perspective, a droplet formation chip could be combined other microchips such as micromixers to create multi-stage microfluidic devices capable of operating at low Re such as an ion transport system57,58. Additionally, further research should evaluate the long-term stability of emulsions produced using this method and investigate how this approach integrates with large-scale manufacturing processes.

Conclusions

For ultrasound emulsification, to achieve uniform droplet sizes and stability of emulsion, high power and multiple cycles are needed. To overcome these limitations, microfluidic techniques can be used as a pre-treatment, where controlled microfluidic flows break the oil phase into smaller droplets. In particular, a microfluidic chip was designed to operate without surfactants, combining the surface properties of the channels interacting with the fluids and a structural design capable of forming droplets. The 3D-designed model was structured so that the needle was positioned at the center of the channel, causing the oil to be disrupted by water in all radial directions. Additionally, a glass capillary with a strong affinity for water was added to guide the oil into a droplet form. As a result, this study demonstrates the use of a microfluidic device to generate uniform oil droplets before the ultrasound emulsification step. By achieving a consistent droplet size initially, the subsequent ultrasound process becomes more efficient, expecting less energy for the emulsification’s stability and precision. This approach offers a viable solution for surfactant-free emulsions through the ultrasonic dispersion equipment.