Passively driven microfluidic device with simple operation in the development of nanolitre droplet assay in nucleic acid detection

Design concept for the simple operation MF-LAMP chip

In the presented device, the system was designed, fabricated and characterized to function as a simple-use microfluidic LAMP chip. The pump-free design and the structure of the MF-LAMP chip are shown in Fig. 1A. An MF-LAMP chip contains 3 channels each with 6 concentric circular chambers in a row. With easy operation, buffers and reagents can be effortlessly accessed and exchanged in the microfluidic channel manually with micropipettes. After reagent injection and oil enclosure, the MF-LAMP chip can be monitored in real-time during the polymerase chain reaction under a fluorescence microscope.

Figure 1B shows the image of a single microfluidic chamber. The concentric circular chamber contains a reaction chamber, a capillary channel and a peripheral channel. The reaction chamber contained approximately 60 nL of reagent in the middle for LAMP observation. Bubble interference is a critical problem, especially in MF systems, but it seems unlikely to influence the stability of the LAMP under a macroscale environment. It can severely affect a microfluidic system, however. Once bubbles form, the air trapped in the reaction chamber expands during the heating process in the LAMP. In addition, the evaporation of the reagent is also a serious problem that leads to changes in both the concentration and volume in large surface-to-volume ratio microfluidic systems.

With the control and regulation of fluid access and preservation, the mechanism is ensured through the geometric structure of the microfluidic design. The basic principle of MF-LAMP operation is illustrated in Fig. 1C. For fluid injection, fluid enters the channel through the buffer inlet, compressing the air equally throughout the entire channel. Additionally, the solution for fluid exchange during the experiment is injected through the buffer inlet. In the enclosed method, as the oil travels through the microfluidic device, the oil stops in front of the capillary channel and bypasses the peripheral channel, causing the remaining reagents to be trapped in the reaction chambers (Fig. 1D and Movie S1). The micropipette as a manual handling tool does not lead to a large error in the droplet volume since the structure is preset in the designed microfluidic system. Hence, bidirectional MF-LAMP is effectively implemented to generate LAMP droplets, avoiding fluid evaporation by simply using micropipettes for operation.

To avoid reagent loss during LAMP thermal heating, oil is applied to the peripheral channel that surrounds the reaction chambers, protecting the middle of the chamber from buffer evaporation. Although the oil enclosure protects the stability of the reagent, the material of the microchannel structure can allow gas dissipation. PDMS, a widely used polymer in microfluidic fabrication, triggers massive gas escape during the heating process. Thin films such as Parylene C, fluorosilane polymer and polyethylene as gas vapor barriers are integrated into the MF chip^42,43,44. However, the process of depositing these materials is complex and time-consuming. In this study, we used a glass slide as the impermeable layer attached to the thin layer of the PDMS microfluidic channel. The impermeable layer binds to the PDMS surface easily directly on the top of the reaction area, avoiding the supporting layer after 1 min of oxygen plasma treatment. In the presence of the impermeable layer and the enclosed oil, the reagent was well preserved in the reaction chamber during the heating process of the LAMP (Fig. 1E). In the absence of both of these features, the molecules inside the reaction chamber evaporate quickly, generating bubbles and condensing the reagent, which severely interferes with the LAMP. In addition, due to their easy access and low cost, glass slides can effectively prevent reduction of the system stability.

Principle of capillary channel control

The capillary channel is a crucial element of the regulation of the passive-driven MF-LAMP device for fluid retainment or passage, as shown in Fig. 2A. The width of the capillary channel directly affects the solution substitution and oil enclosure efficiency in this system. The preparation process requires buffer washing and reagent displacement from the buffer inlet, and the solution exchange efficiency is strongly dependent on the width of the capillary channel (Fig. 2B). The narrow capillary channel limits the fluid velocity in the reaction chamber, resulting in a low exchange efficiency for the buffer. In the simulation of the buffer exchange efficiency for capillary channels of different widths, the flow characteristics were solved by using the simplified Navier-Stoke equation and continuity equation:

where u is the fluid velocity, ∇ represents the gradient, p is the fluid pressure, ρ is the fluid density, μ is the fluid dynamic viscosity, F is the external force applied to the fluid, I is the identity matrix and T is the matrix transpose function. The wall condition was set to no slip. Additionally, the convection and diffusion effects in the mass transport for the concentration gradients were considered in the solution exchange simulation.

where N_i is the transport flux of the species, ci is the species concentration, Di is the species diffusivity and u is the fluid velocity.

In the initial stage (0–0.5 s), the channel is prefilled with one solution (shown in blue), while the other solution, shown in red, enters the buffer inlet subsequently at 0.5 s with a flow rate of 0.6 mL min⁻¹. The mixing process is presented in Fig. 2C with channel widths of 10, 30 and 50 μm. The wider capillary channel shows a higher exchange rate with a higher velocity in the reaction chamber. The solution substitution ratio for both the 30 and 50 μm channel widths reached 100% within 2.5 s. In contrast, the 10 μm channel requires a longer substitution time due to the slow fluid velocity in the reaction chamber, which leads to increased reagent consumption. The solution substitution efficiency demonstrated that within 1 s, the 30- and 50-μm channels achieved 100% solution replacement. However, the narrow-width 10 μm channel requires a longer time (> 2.5 s) for solution substitution (Fig. 2D). To better clarify the velocity distribution, the velocity profiles across the Y-axis in the reaction chamber are shown in Fig. S2.

Compared to the peripheral area, the central areas of the reaction chamber showed a higher velocity that corresponded to the solution exchange process, with the substituted solution observed first in the center and then expanding into the peripheral area. It also shows the velocity decrease with the decrease in the width of the capillary channel. To test the performance in actual operation for solution exchange, distilled water was preinjected into the reaction chamber. With a micropipette, 1 µg mL⁻¹ fluorescein in distilled water was injected into the channel (Fig. 2E); the fluorescence intensity response was recorded, and the exchange efficiency was calculated. Within 2.5 s, the fluorescence solution was 100% substituted by the distilled water, which was similar to the simulation result (Fig. 2F).

In the process of sample loading, formation of stationary droplets with stable volume and conditions is necessary for a LAMP droplet assay. Additionally, oil enclosure was adopted to ensure a more reliable process by decreasing the water loss in the reaction chamber during thermal heating. As the mineral oil enters the microchannel from the opposite direction of the buffer inlet (Fig. 3A), it travels along the microfluidic system and is stopped by the capillary channel. This capillary barrier is achieved by abrupt expansion of different sections of the wettable microchannels that break up the aqueous reagent, generating uniform droplets depending on the microfluidic geometry⁴⁵. The Young–Laplace equation describes the bypass pressure in a rectangular microchannel as

where w and h are the width and height of the microchannel, and σ represents the surface tension of the liquid. The advancing contact angle of the sidewall is θ_s, and that of the top and bottom is θ_v. The pressure barrier of the fluid expansion in this case depends on the geometry of the channel, while the surface tension and contact angles remain the same. By the time the oil reached the capillary channel, as long as the insertion pressure did not exceed the bursting pressure, the oil tended to bypass the reaction chamber and flow through the outer channel (Fig. 3B), where it was divided into uniform droplets with oil protection. As determined by the numerical simulation, Fig. 3C illustrates the simulated domain with different capillary channel widths (10, 30 and 50 μm). The contact angle of the oil with PDMS in the aqueous phase was set as 120°. During injection at 5 mm s⁻¹, the advancing oil was stopped at the 10- and 30-μm capillary channels, preserving the LAMP reagents in the reaction chamber. In contrast, the 50 μm capillary channel had a lower bursting pressure, which was the result of increasing the width in Eq. (5), and the insertion pressure easily reached the threshold value. The simulated illustration shows that the fluid in the reaction chamber was pushed out during the oil enclosure, with reagent loss (Fig. 3D). The actual oil loading process is shown in Fig. 3E, and blue dye was used to provide better visualization of the oil enclosed process. The reagent volume variation is calculated by the image j of the ratio of V_oil and V_i, where the initial volume in the central reaction chamber is V_i and the reagent volume after oil injection is V_oil. The reagent was 100% in volume, which remained in the reaction chamber during the following reaction and observation (Fig. 3F). Oil insertion not only divided reagents into uniform droplets but also prevented evaporation during the heating process.

Based on the simulation and actual testing results, we suggest the proper design of the geometry of the MF-LAMP chamber. Despite the fact that the narrow width is more appropriate for oil-enclosed regulation of the bursting pressure threshold, the velocity in the reaction chamber is quite low, resulting in low solution substitution efficiency.

Characterization of the MF-LAMP device

Maintaining the proper concentration is vital for quantification LAMP method. The slight differences in samples, primers, nucleotides, ions, buffers and temperature strongly affect the function and efficiency of polymerase replication. To test the stability of the reagent, fluorescent dye was injected into the reaction chamber with enclosed oil. Figure 4A illustrates the different chip methods, including the MF-LAMP method (with oil enclosure) and enclosed heating method without oil at 70 °C for 30 min. The initial and after heating images depict the effect of different strategies of chip fabrication, and blue dye and fluorescent dye were inserted into the microfluidic device for bright field microscopy and fluorescence image acquisition, respectively. The variation between the initial status and the status after heating for 30 min for the fluorescence intensity is shown in Fig. 4B. In the MF-LAMP chip, the solution maintained a well-preserved condition with no bubble interference in the reaction. The relative standard deviation was measured as 1.8% in the initial state and 4.8% after 30 min of the heating process.

The outer peripheral channel could also be the buffer area for liquid expansion during increases in heat. In contrast, the chip without enclosed oil showed great variation after the heating process. Without oil protection, evaporation easily occurs in the inlet adjacent channel, with massive water loss. Bubbles also invaded both ends of the reaction chamber and even squeezed out the reagent inside the chamber, resulting in difficulties in fluorescence detection. Since there was no oil to divide the reagent into uniform droplets, the evaporation also led to condensation of the reagent in the middle area, resulting in high fluorescence intensity and great variation in the RSD value (1.4% in the initial state and 36.4% after thermal heating). Additionally, in the fabrication of the MF-LAMP chip, we also considered the gas/liquid permeability of PDMS during the heating process.

The functional, impermeable layer was adherent to the high porosity PDMS as a vertical gas/liquid barrier that lowered the diffusion of solution. In comparison to the necessity of each functional process or layer, different methods, including the MF-LAMP method (with both oil enclosure and the impermeable layer) and methods without oil enclosure or an impermeable layer after heating at 70 °C for 30 min, are shown in Fig. S3. With high porosity, the PDMS polymer provides “openings”, allowing gas molecules to diffuse inside the networks. Especially under thermal heating conditions, the transport of gas/liquid molecules through PDMS increases due to the high pressure gradient between the reaction chamber and the external environment. Such an increased mass flow rate generates water dissipation and condenses the reagent inside the reaction chamber, resulting in high fluorescent intensity both in the fluorescence images and the trend graph. The RSD over time for MF-LAMP is 2.0%, which is relatively lower than that observed without oil enclosure (10.1%) and without an impermeable layer (11.5%). Consequently, no bubble invasion or reagent loss occurred in the MF-LAMP chip during thermal heating. Both results showed that our approaches for MF-LAMP chips are effective in suppressing bubble formation and solution condensation.

Application of the MF-LAMP device for diagnosis

For the MF-LAMP assay, the droplet assay was performed under oil with a uniform composition depending on the geometric structure of the microfluidic design. To evaluate the performance of droplet manipulation under manual operation conditions, we generated a 3 × 6 assay of a single droplet filled with various fluorescence dye concentrations of 0.1875, 0.375, 0.75, 1.5 and 3 μg mL⁻¹ (Fig. 4C). The fluorescence intensities in each assay were measured, and the relative standard deviations (RSDs) were calculated as 15.6%, 3.2%, 5.0%, 1.0% and 2.0% (n = 18). Under a low fluorescence concentration, the measured RSD was relatively high, and the weak intensity was easily affected by the background noise. On the other hand, in the range from 0.375 to 3 μg mL⁻¹, a low RSD resulted, which showed the high precision and reliability of the MF-LAMP device with uniform droplet manipulation under manual operation mode.

To evaluate the usability and sensitivity of the MF-LAMP device for nucleic acid amplification application, Escherichia coli (E. coli) malB gene samples with a wide range of concentrations (from 1 ng μL⁻¹ to 1 pg μL⁻¹) in the droplet assay were measured. The target samples were mixed with LAMP primers and reagents and inserted into the MF-LAMP chip, followed by oil enclosure for channel sealing for the heating process. The fluorescent images of droplets were recorded, and the corresponding real-time intensity variation curves are presented in Fig. 4D as the linear fitted graph of the logarithm of the concentration of malB template versus the reaction time, and the linear relationship was obtained (R² = 0.94). The standard deviations of the reaction times were 0.645, 0.545, 0.288 and 0.328 for concentrations of 1 ng μL⁻¹, 100 pg μL⁻¹, 10 pg μL⁻¹ and 1 pg μL⁻¹ (n = 6), respectively. The gel electrophoresis after reaction was showed in Fig. S4. The result showed that there is difference in 1 pg μL⁻¹ and primer only solution. The proper incubation time should adjust to 30 to 40 min by adjustment of the copies/ml in the clinical sample. For this reason, we adjust the LAMP reaction time within 60 min and suggesting the proper quantification time around 30 to 40 min. It makes sure that the signal amplification that they are seeing is not coming from primer dimerization. In addition to gene detection, we also demonstrated the mRNA expression of the mutant epidermal growth factor receptor (EGFR) gene in H1975 cell lines on the MF-LAMP chip, and the results are shown in Fig. S5. These results demonstrate the uniformity and feasibility of the droplet assay for both gene detection and mRNA expression detection by using the nucleic acid amplification microfluidic technique under manual operation with simple steps. Additionally, because of its high sensitivity of detection, MF-LAMP also substantially reduces sample and reagent consumption.