Droplet microfluidic chip for precise monitoring of dynamic solution changes

The homeostasis and dynamics of intracellular calcium ions play a crucial role in neural communication in the brain. To better understand brain function, it is essential to accurately monitor changes in its concentration within neurons. Detecting fluorescence, using probes such as fluo-4, is commonly used for monitoring calcium ions. However, the high temporal resolution required for monitoring such rapid dynamics remains a challenge.

However, when monitoring biochemical molecules, collecting, manipulating, and conducting a high temporal resolution analysis of nanoliter samples in vivo is still challenging.¹ Sampling and monitoring rapidly changing biochemical signals in a tiny liquid sample with a high temporal resolution are significant for studying kinetic reactions.² A widely used liquid biopsy method in neuroscience for quantifying the release of neurotransmitters is microdialysis. Its advantages include the small size of the equipment, minor trauma, and long-time continuous monitoring of drug concentrations.³ However, the primary concentration of samples collected by microdialysis probes needs to be revised by recovery, which increases the complexity of the experiment and limits the accuracy of the results. It also has a low temporal resolution of up to minutes because of the low perfusion rate and Taylor dispersion.⁴ 

One way to reduce the size of monitoring devices is microfabrication. There is less tissue trauma due to the high-resolution sampling microstructure.⁵ Electrochemical probes based on microelectrodes can achieve high spatial resolutions down to 0.003 mm² and time resolutions down to several seconds for in vivo monitoring.⁶ They have a low threshold and can simultaneously detect glutamate below 500 nM and dopamine below 20 nM.⁷ However, they rely on selective enzymatic reactions to detect a specific molecule, so the devices must be calibrated regularly.⁸ 

Droplet microfluidic chips are a widely used and promising technology for precisely manipulating and analyzing small volumes of fluid,⁹ typically from nanoliters to microliters.¹⁰ The solution is encapsulated in oil. These chips are considered to be a powerful tool in biomedical research with advantages such as small independent reactors, high throughput, minimal Taylor diffusion,¹¹ less channel surface absorption, and effective mixing.¹² Droplets can also be ideal containers for biopsy sampling¹³ because they reduce the impact on the target in vivo due to the small sample volumes.¹⁴ Isolated samples can be collected at different times. The devices have high temporal¹⁵ and spatial resolutions in situ or in vivo.¹⁶ 

Many kinds of backend biomolecules have been used for accurate, efficient, and high-resolution online detection of droplets,¹⁷ based on techniques such as fluorescence,¹⁸ chromatography, mass spectrometry,¹⁹ electrochemistry, and conductivity.²⁰ 

Researchers have explored the use of droplet microfluidic chips for monitoring dynamic changes in solutions in various fields.¹² For example, one application is measuring chemical kinetics. Droplet microfluidic chips have been used to study the kinetics of chemical reactions in real time with high precision.²¹ Another application is in environmental monitoring, as microfluidic chips can detect changes in the concentration of pollutants in water over time. Additionally, droplet microfluidic chips have shown promise in diagnostics, because they can detect biomarkers in bodily fluids and provide results in minutes.²² 

Researchers have recently focused on developing droplet microfluidic chips for monitoring dynamic changes in solutions in real time.²³ The continuous monitoring of a solution requires the integration of sensors and other components into a microfluidic chip.²³ For example, researchers have developed droplet microfluidic chips that can monitor changes in pH and temperature during reactions under controlled conditions. Other researchers have developed microfluidic chips that can monitor changes in the concentration of analytes over time, which may be a powerful tool for environmental monitoring and biomedical research.²⁴ 

Because of these advantages of droplet microfluidics, many researchers have made progress in bioscience applications. In Song’s work, nanoliter droplets served as reaction tanks when measuring the millisecond kinetic parameters of an enzymatic reaction.²⁵ They were able to determine reaction times precisely by measuring the transport distance. However, regrettably, injecting pretreated samples into a channel and generating droplets is inappropriate for in vivo collection. So, Feng developed a microfluidic needle for sampling and delivering chemical signals by segmented flows.²⁶ The method produced droplets with good volumes and frequency consistency when determining the concentration of hydrogen peroxide. A push-pull perfusion microprobe was developed by Van den Brink et al. It can generate droplets for sampling and online detection with high chemical and spatial resolutions and a temporal resolution of 3 s.⁶ It is a useful tool for fundamental neuroscience research, but the temporal resolution of 3 s still needs improvement because the neurotransmitter release process takes only milliseconds. Li et al. developed an automated droplet generation and analysis device based on a push-up valve with high resolution when sampling and measuring hormone secretion.²⁷ 

Overall, the use of droplet microfluidic chips for monitoring dynamic changes in solutions has the potential to revolutionize a wide range of fields, from drug discovery and environmental monitoring to point-of-care diagnostics.^28,29 As researchers continue developing and refining this technology, we expect to see many exciting new applications.

This work explores the potential of droplet microfluidic chips for monitoring applications. The workflow of our device is illustrated in Fig. 1. In summary, a negative pressure pump pulls the fluid through the chip. The fluorescence is recorded on video using a microscope and used for flow analysis.

The device utilizes negative pressure to transport liquid samples into a T-shaped channel where the detection reagent is added. A flow-focusing structure is then used to generate sub-nanoliter droplets. These are mixed in the sample incubation area before being transported to the detection area below the microscope. This approach enables real-time monitoring of rapidly changing molecules and is an efficient online fluorescence monitoring platform.

The materials used in this study included a negative photoresist (SU-8 3050, MicroChem, Newton, MA) and 4-inch silicon wafers (MCL Electronic Materials). Lithography was performed using a maskless aligner (MicroWriter ML3, Durham Magneto Optics Ltd.). Polydimethylsiloxane (PDMS; Sylgard 184) was used as the material for the microfluidic chip. Fluorescent calcium indicator (Fluo-4, Thermo Fisher) and fluorescein dye (Sigma-Aldrich) were prepared in different concentrations. A plate reader was used for detection (Victor 2 V Multilabel Counter, PerkinElmer). The oil used in droplet formation was homogenized with surfactants using an ultrasonic bath (Selecta Ultrasonic). A 10 mM solution of CaCl₂ in 20 ml of deionized water was prepared as the calcium ion source and lower concentrations were prepared using serial dilution.

This microfluidic chip was fabricated by a soft lithography process, as shown in Fig. 2. The chip consists of a glass slide and a PDMS block. It has three inlets for the sample, reagent, and oil, which are produced by punching holes into the ends of the channels. A pipette tip can be plugged into each of these holes.

The process flow is illustrated in Fig. 2(a). Figure 2(b) further demonstrates the microchannel fabrication and punching process as well as the hierarchy of the chip. Soft lithography was used to fabricate the microfluidic chip from PDMS as follows. A mixture of PDMS was poured into a mold made from the photoresist. It was then heated and cured to cross-link it with the sample and finally stripped off. The PDMS and glass were treated with oxygen plasma and placed together so that covalent bonds formed between the PDMS and the glass sheet. The PDMS was then heated in an oven at 110 °C for 12 h to restore its hydrophobicity so that stable droplets could be generated.

A vacuum pump with a negative pressure valve was connected to the chip outlet and used to drive the flow. The mixing process at the front of the chip is depicted in Fig. 3. First, mineral oil is injected into the chip and dye solution is injected into the reagent channel. Negative pressure of −10 kPa then extracted the air in the chip. Since the pressure of the reagent channel is less than that of the sampling channel, the reagent flowed into the droplet generation point. The transparent sample flowed into the channel where it immediately mixed with the dye solution and the mixture separated into droplets. Then the negative pressure was increased to −15 kPa to drive the sample reagent to discharge the oil and dye solution into the sampling channel. If the liquid flow of the two aqueous phases is stable, droplets naturally form which continue to mix or react during their transport along the channel. The droplets were mixed entirely with the reagent solution during transportation and detected by fluorescence microscopy as they flowed through the channel.

The channel size controlled the size and speed of the droplets. The length of the droplets could be fixed by adjusting the pressure valve. Thus, the process in the chip is stable. The chip can be seen as an analog-to-digital converter because it transforms the continuous concentration of the solution into discrete droplets arranged in sequence at different positions in the channel. This experiment verified the ability of the chip to collect and mix samples with a reagent and form droplets.

The statistical results indicate that droplets were generated with a frequency of 42 ± 2 Hz under a −15 kPa negative pressure. The droplet length was consistent with an average value of 223 ± 11 μm in a 200 μm width channel, indicating that the volume of each droplet was about 4.19 nl and that the sample flow rare was 88 nl/s. These results suggest that a microfluidic droplet chip can effectively monitor the dynamic changes of a solution.

Two kinds of solutions come into contact when they reach the T-junction in the microfluidic droplet chip. They were efficiently mixed during droplet transportation, resulting in consistent and accurate immunoassay results, as shown in Fig. 3. The mixing ratio was determined through fluorescence imaging of droplets containing fluorescein mixed with a buffer. Characterization of 33 droplets of each concentration revealed that the fluorescent intensity of the droplets was 48.05 ± 0.22% in the reference experiment, indicating a mixing ratio of 1:1.08. The mixing precision was high, with a relative standard deviation of less than 0.5%. This consistent mixing ratio for the sample solution and probe solution within each droplet is crucial for accurate droplet immunoassays. Thus, this chip is a reliable platform for monitoring dynamic changes in a solution.

A mixer with two reagent channels can be used for more complex reactions with different combinations of reagents. However, when there are two reagent channels, it is difficult to ensure that both phase reagents are stably mixed with the sample at the same time. Thus, the stability of the chip is reduced, making the subsequent analysis more difficult. Therefore, a feasible approach is to pre-mix multiple reagents off-chip and then mix them with the sample in a single reagent channel.

Our aim is to put as much of the droplet generation and mixing structure as possible at the back end of the chip. This would reduce the size of the front of the chip, allowing it to be developed into a probe for in vivo detection in the future. Moreover, we want to keep the test reagents some distance from the sampling point to avoid mutual contamination. This would prevent the fusion or interference of droplets in the reaction channel after mixing. Because detecting fluorescence needs time to incubate, long exposures are needed to collect the weak fluorescence signals.

Thus, we designed two microstructures for generating, mixing, and capturing droplets (Fig. 4). The droplet generation structure of chip A is shaped like a needle tip. This structure is followed by a fusion trigger structure and a trapping structure. Chip B has reagent inlets before and after the droplet generation structure. This is followed by a fusion trigger structure, a zigzag incubation channel, and a different type of trapping structure. However, our experiments found that, although it was easy to determine the volume of the mixed reagent and clarify the specific reaction time point for the structure in which the detection reagent was mixed into a droplet after the droplet had been produced,³⁰ the mixing was unstable and not all droplets were mixed properly.

The most significant difficulty for the droplet capture structure is ensuring the sequential arrangement of droplets, which this structure cannot accomplish. Thus, we adopted a bypass structure. However, the bypass structure requires that the channel is accurately designed and manufactured and also needs a high driving pressure. Therefore, we finally found that the best structure is the continuous zigzag microchannel, which has a simple design and a high area utilization rate.²⁹ Most importantly, it generated the best sequential arrangement of droplets, so we chose this structure.

We tested the temporal and concentration resolutions of the chip by monitoring a rapidly changing fluorescent dye solution. To do this, we manually adjusted the solution within the reservoir. We tested the assay performance by quickly changing the solution concentration from 1 to 5 μM using a pipette at the sample inlet. A high-concentration dye solution was added to the existing low-concentration solution so that each subsequent droplet had a higher concentration.

We recorded a video of the droplets flowing through the downstream chip channel under the objective lens of a fluorescence microscope camera. However, the high flow speed of the droplets limited the exposure time, necessitating a higher fluorescence intensity than for static objects. We resolved this problem by improving the intensity of the excitation LED light source (100%), maximizing the camera amplification (iso1600), using the highest magnification possible (10×), and merging a 4 × 4 pixel array into 1 pixel.

The video was analyzed with the software ImageJ, which extracted the fluorescence intensity over time for a selected line across the channel. We then processed the data using MATLAB to extract the top edge of the point clusters and smooth the discrete points into a plane curve. Figure 5(a) shows the mean gray values varying rapidly with time, indicating the ability of the chip to sample and detect chemical changes. The droplet microfluidic chip showed high temporal and chemical resolutions, enabling the rapid detection of changes in fluorescence intensity when a higher concentration solution was added to an existing solution. The small dead volume and high-frequency of generating droplets contributed to the high resolution of the chip. The drop observed following a peak was due to dilution. The chip has an excellent temporal resolution of 3–6 s and a chemical resolution of 1 μM for fluorescent dye at a droplet flow frequency of 42 Hz. These findings demonstrate the ability of the droplet microfluidic chip to detect concentration changes that take more than 3 s. This remarkable result surpasses the capabilities of existing similar devices and needles.

Moreover, we measured the fluorescence intensity for different concentrations of the dye solution using a 384-well plate and a plate reader. As illustrated in Fig. 5(b), the fluorescence intensity of the droplets had an excellent linear relation with the concentration. We obtained similar results when detecting droplets generated by the chip and recorded on video. We verified these results after storing the droplets in the channel for 5 min. These results confirm that the chip can convert high-speed flow detection into static detection using the space-for-time method, as dynamic changes are recorded in the discrete droplets.

We measured the diameter of the droplets after mixing within 5 min to judge the sampling and the volume after mixing, and found that the change in the diameter of the droplets after mixing was about 5% in the range of 222 μm, indicating that the difference between the volume of each droplet in this sampling and the mixing method was small, laying the foundation for stable sampling and detection, and avoiding the influence of droplet volume change on detection [Fig. 6(a)]. The decay of fluorescence was compared for two methods: long-lasting continuous exposure and short-time-interval exposure (10 ms) in each minute,, which simulates multi-detection for flowing droplets [Fig. 6(b)]. Pictures were taken at the same time intervals over 8 min under the two conditions. Continuous exposure led to a 50% decay in fluorescence, whereas the interval exposure resulted in only a 10% decay. Each flowing droplet was exposed for less than a second. This suggests that for multi-detection applications, the decay of the fluorescence for flowing droplets is reduced by ∼40 percentage points. This finding highlights the potential of using flowing droplets to reduce interference and could improve the accuracy and reliability of back-end fluorescence measurements of a collected sample.

We mixed 10 μM of fluo-4, a fluorescent calcium indicator, with CaCl₂ solution in a one-to-nine ratio in 96-well plates. As measured by a plate reader, the fluorescence intensity had an excellent log-linear relation with CaCl₂ concentration in solutions ranging from 100 nM to 10 mM (Fig. 7). Thus, we chose this concentration range for the experiments with our chip.

For future work, we will determine the linear range and threshold for detecting calcium ions with our device. As above, the liquid sample will be introduced into the sampling channels, where the sample will mix with a reagent to generate fluorescent material. It will then enter a flow-focusing structure with oil so that the aqueous phase will separate into droplets. We plan to mix different concentrations of calcium ions with detection reagents outside the chip, then detect the concentration inside the chip to obtain a standard curve for detecting calcium ions. Thus, we will realize in-chip mixed detection for different concentrations of calcium ions in the bulk reservoir. We hope this method can be used to detect calcium ions in real samples, and we plan to extend this method to more targets.

In this study, a droplet microfluidic chip was developed. Droplets are generated and driven under negative pressure to meet the requirements of future in vivo sampling applications. This approach improves the detection resolution, which is limited by Taylor dispersion, by encapsulating samples in discrete droplets. The chip could extract sub-nanoliter samples from the front end and mix them with a detecting reagent to produce droplets that a fluorescence microscope can detect. It can accurately measure changes in concentration with a high temporal resolution. This study investigated the relations between sampling frequency, applied pressure, and mixing ratio and tested the capability of the chip as a sampling device. We demonstrated its linear range by sampling a series of diluted fluorescein solutions from 10² to 10⁷ nM. The droplets were generated at a frequency of 42 Hz under −15 kPa with a mixing ratio of 1:1. The temporal resolution was 3–6 s. Our results demonstrate the feasibility of using a droplet microfluidic chip for dynamically sampling the concentration of a solution by detecting fluorescence. Its advantages over microdialysis probes include high temporal resolution, smaller sampling volume, and controllable mixing ratio.

We established that the detection limit of fluo-4, a calcium ion indicator, was 10 nM, thus laying a foundation for detecting low concentrations in vivo. Our future work will focus on practical applications, such as monitoring the concentration of calcium ions released by cells in a culture medium and the secretion of other small molecules using fluorescence-based probes. We will assess its capability to dynamically measure changes in the calcium concentration of a solution by mixing the sample with fluo-4 on-chip. Moreover, we are trying to use this chip with an implantable silicon-based microneedle, as we mentioned in our previous report.³¹ 

Overall, the droplet microfluidic chip developed in this study is a promising platform for high-resolution dynamic monitoring of changes in the concentration of a solution.

Despite its advantages, this microfluidic chip still has several limitations. First, the dynamic range is limited to the 255 quantitative levels measured by the 8-bit camera, so that it can saturate during constant exposure if the fluorescent intensity increases significantly. The dynamic range could be expanded four times using a 10-bit camera or by normalizing the fluorescence intensity for different exposure parameters, such as using a long exposure time for lower-concentration solutions and a shorter time for higher-concentration solutions. Second, the low sensitivity of the microscope camera limits the detection frequency compared to a photomultiplier tube. A higher detection sensitivity would require a shorter exposure time and allow the flow rate to be increased. Third, the large size of the chip currently limits its use for in vivo sampling. However, a specially designed cutting process could be used to create a sharp needle tip, which could reduce trauma but may also cause a lack of rigidity during implantation. The platform may, instead, be fabricated with a silicon-based needle, as described in the literature.

We acknowledge support from the equipment research and development projects of the Chinese Academy of Sciences, “On-chip integrated optical biochemical detection key technology research and development team,” E11YTB1001. We also thank our research platform: the State Key Lab of Transducer Technology at Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences.

The authors have no conflicts to disclose.

C.M. and Z.G. are co-first authors of the article.

The first author, Cong Ma, was responsible for designing, fabricating, and testing the microfluidic chip, for making measurements, and for writing this paper. Zehang Gao helped with experiments and revised the final manuscript. Corresponding authors Jianlong Zhao and Feng Shilun provided the experimental environment and helped write the manuscript.

The data that support the findings of this study are available within the article.

Cong Ma is focusing on design and fabrication of microsensor and it backend circuits in Shanghai Institute of Microsystem and Information Technology, Chinese Academy Sciences. He is a master’s degree candidate at ShanghaiTech University, where he majors in Electronics Science and Technology.

Zehang Gao is a post-doctor in Department of Clinical Laboratory Medicine, The Third Affiliated Hospital of Guangzhou Medical University, China. His current research is focused on droplet-based microfluidics.

Shilun Feng is focusing on different point-of-care testing (POCT) researches for food, environmental water and biomedical sensing. He is currently an associate professor in State Key Laboratory of Transducer Technology, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai. He just finished his Research Fellow journey for POCT microfluidics projects on environmental water in School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. He completed his Ph.D. in Biomedical POCT microfluidics with Dr. David Inglis, specialised in POCT microfluidic sampling probe and POCT on-chip cell concentrator, in the School of Engineering, Macquarie University, Australia. His research interests include biomedical microfluidics; microfabrication; point-of-care (POC) sampling, manipulation and testing with the developments of biodevice and instrumentation systems.

Jianlong Zhao received his bachelor’s degree from the Department of Electronic Engineering, Tsinghua University in 1992 and his doctor’s degree from Shanghai Institute of Metallurgy, Chinese Academy of Sciences in 1997. Currently, he is the deputy director, researcher and doctoral supervisor of Shanghai Institute of Microsystems, Chinese Academy of Sciences, deputy director of the academic committee of Shanghai Institute of Microsystems, executive deputy director of Shanghai Wireless Communication Research Center, and deputy director of Shanghai National Engineering Center for Biochips. He is mainly engaged in the research of biochips and biosensors, and has been responsible for many important scientific and technological projects such as major projects of CAS, major projects of Shanghai Science and Technology Commission, and national 863 program. He was the chief scientist of the major project “Research and Application of Biochip System” of the Knowledge Innovation Project of CAS.