Luer-lock valve: A pre-fabricated pneumatic valve for 3D printed microfluidic automation

Abstract

3D printed microfluidic devices are made of stiff and easy-to-fatigue materials and hence are difficult to have robust pneumatic valves. In this work, we describe a type of prefabricated polydimethylsiloxane (PDMS) valves, named the “Luer-lock” valve, which can be incorporated in 3D printed microfluidic devices utilizing the Luer-lock mechanism. Luer-lock design has been adopted for fluidic connections worldwide; it is facile, reliable, and inexpensive. To take advantage of the Luer-lock design, we added “valve ports” to our 3D printed microfluidic devices; prefabricated PDMS valve modules could be embedded into these valve ports, in a leak-free manner, by screwing tight the Luer-locks. In the experiment, we succeeded in fabricating pneumatic valves with a footprint diameter of 0.8 mm and verified the functionality of these valves with a shut-off pressure of 140 mbar and a maximal switching frequency of ∼1 Hz. As a demonstration, we show the serial encoding of core–shell hydrogel microfibers using the Luer-lock valves. Since the Luer-lock valves can be mass-produced and the CAD model of Luer-locks can be easily distributed, we believe that our approach has the potential to be easily adopted by researchers around the globe.

I. INTRODUCTION

3D printing, since its emergence in the 1980s, has been continuously revolutionizing the traditional manufacturing industries. Recently, with its drastically improved resolution (up to hundreds of nanometers) and flexibility in material choices,^1–6 3D printing is also penetrating through various microfabrication-related fields, such as microfluidics. Using high-resolution single-material printing techniques, microfluidic devices can be rapidly prototyped with modular designs,⁷ easy world-to-chip connections,^8,9 complex 3D channels,^10–13 and also automation functions based on pneumatic membrane valves.^14–18 Since the single-material valve membranes, which are generally made of stiff plastic for high printing resolution, are often operated to deform non-linearly at a distance much larger than their membrane thickness and, therefore, easy-to-fatigue, recently, multi-material valves have been proposed by incorporating elastomeric membranes into the 3D printed devices.^19–22 However, these methods require the assembly of multi-materials using complicated printing procedures, chemical bonding, or clumsy clamping mechanisms to achieve leak-free sealing after the assembly.

In this work, we propose a type of “Luer-lock” valve—an elastomeric valve that is easy-to-configure in 3D. The valve modules are prefabricated using polydimethylsiloxane (PDMS) molding and can be easily embedded into 3D printed microfluidic devices by screwing tight the Luer-lock connectors. The high flexibility of PDMS (with a reported Poisson's ratio as high as 0.5)²³ makes the valves highly durable. The valves are also disposable and field-deployable since they can be mass-produced and can be easily installed and uninstalled manually; once failed, these valves can be easily repaired or replaced to avoid the abandonment of the whole microfluidic device. In experiments, we validate the design and functionality of these valves (footprint diameter: 0.8 mm, membrane thickness: ∼100 μm) with a shut-off pressure of 140 mbar and a maximal switching frequency of ∼1 Hz. As an application, the Luer-lock valves can be integrated into a 3D printed double coaxial microfluidic device to enable automatic fluid switching functions; using this device, we demonstrate the fabrication of serially encoded core–shell hydrogel microfibers, which are potentially useful for tissue engineering and regenerative medicine purposes.

II. CONCEPT AND DESIGN OF THE LUER-LOCK VALVES

A 3D printed device with a Luer-lock valve is designed to be assembled from three parts: a 3D printed device with a “valve port,” an elastomeric “valve module,” and a commercially available male-type Luer-to-barbed fitting. First, as shown in Fig. 1(a), the 3D printed device has to be specifically designed with a “valve port.” At the valve port, the micro-channel, which will be controlled by the Luer-lock valve, is opened with a small window to its upper side and docked in a female-type Luer connector. Second, the “valve module” is a PDMS hollow cylinder with a thin PDMS membrane bonded on its bottom [Fig. 1(b)]. The valve module is a type of prefabricated valve²⁴ that can be mass-produced using PDMS molding: an array of hollow cylinders can be made by molding PDMS, then be bonded with a spin-coated PDMS membrane using oxygen plasma treatment, and finally be diced into separate valve modules. Third, a male-type Luer fitting, which converts the Luer-lock to a barbed connector, is a commercial product that can be easily purchased worldwide. As shown in Fig. 1(b), the three parts are assembled as follows: first, a valve module is deployed into the valve port, with the bottom PDMS membrane of the valve module capping onto the open window of the microfluidic channel; then, a male-type Luer-to-barbed fitting is applied from above and screwed tight to cause a slight deformation of the PDMS elastomer, which not only seals the microfluidic channels in a water-tight fashion but also guaranteed gas-tight connection to the pneumatic source through barbed connectors. During the valving operation, the control pressures P_C will be applied through the barbed connector to alter the deformation of the bottom PDMS membrane; when the pressure applied is higher than a threshold pressure P_threshold, the microfluidic channel beneath the membrane will be closed, thus realizing the valving function. Alternatively, when the valving function is not required, one could also seal the valve port using commercially available Luer-lock stoppers, which will end up with a normal microfluidic channel and, therefore, increase the flexibility of the fluidic channel network configuration.

FIG. 1.

Concept and design of the Luer-lock valves. (a) A representative 3D printed microfluidic channel, which can be configured with the Luer-lock valve at the “valve port” featured with a female-type Luer design for easy valve configuration. After valve configuration, the on/off of the flow, from the inlet to the outlet, can be controlled by the application of a waveform of control signals (i.e., the pneumatic pressures). (b) The configuration process and working principle of the Luer-lock valve illustrated under section view: first, a valve module is deployed into a valve port, with the bottom PDMS membrane of the valve module capping onto the open window of the microfluidic channel of the valve port; then, a Luer fitting is applied from above and screwed tight to cause a slight deformation of the PDMS elastomer, which not only seals the microfluidic channels in a water-tight fashion but also guarantees gas-tight connection of the pneumatic source through barbed connectors. During the valving operation, control pressures P_C will be applied through the barbed connector to control the deformation of the bottom PDMS membrane; if the pressure applied is higher than a threshold pressure P_threshold, the microfluidic channel beneath the membrane will be closed. Scale bars: (b) 6 mm.

III. RESULTS AND DISCUSSION

A. Design and fabrication of the valve modules

For the fabrication of the valve modules, step-by-step illustrations of the process under a section view are shown in Fig. 2(a). The steps of fabrication can be separated into three parts: part A, part B, and part C.

FIG. 2.

Fabrication of the prefabricated PDMS valve modules. (a) The step-by-step illustration of the fabrication process of the valve modules under section view. Briefly, the steps of fabrication can be separated into three parts. In part A, a thin film of PDMS with a thickness of ∼100 μm is achieved using spin-coating. In part B, an array of ring-shaped PDMS structures was produced in a single batch of PDMS molding using a polymethyl methacrylate (PMMA) mold. In part C, the thin PDMS film (on a silicon wafer) fabricated in part A and the array of ring-shaped PDMS structures fabricated in part B were bonded together and diced into valve modules. (b) PMMA mold used in part B. (c) Photograph of a valve module with an inner diameter of 1.5 mm and an outer diameter of 4 mm. (d) Multiple valve modules can be fabricated in a single batch.

In part A, the PDMS precursor mixture (two-component, mixed at a 10:1 ratio, SYLGARD™ 184 Silicone Elastomer Kit, The Dow Chemical Company) was gently poured onto a silicon wafer and spin-coated with a gradually ramping-up rate from 0 to 1000 rpm (plateaued for 30 s); a thin film of PDMS with a thickness of ∼100 μm can be achieved after baking the spin-coated wafer on a hotplate at 75 °C for 1.5 h. Note that before pouring the PDMS, the silicon wafer was coated using 5 wt. % polyvinyl alcohol [PVA, molecular weight (MW) 13 000−23 000, 87%–89% hydrolyzed, 363170, Sigma-Aldrich] aqueous solution under 1000 rpm for 30 s, followed by baking the wafer on a hotplate at 85 °C for 15 min. The wafer was cooled down to RT by blowing nitrogen gas on the backside of the wafer before spin-coating the PDMS film. The PVA coating provides a water-soluble intermediate layer, which will facilitate the easy peel-off of the PDMS thin film from the silicon wafer.^25,26

In part B, an array of ring-shaped PDMS structures was produced in a single batch of PDMS molding with a polymethyl methacrylate (PMMA) mold [Fig. 2(b)]. The PMMA mold was fabricated using a programmable milling machine (MM-100, Modia Systems Co., LTD) and subsequently coated with Parylene-C (2 μm). Then, the PDMS precursor mixture (same as used in part A) was poured into the mold, de-bubbled in a vacuum, capped using a thin polyethylene terephthalate (PET) film, and cured on a hotplate at 65 °C for 2 h followed by RT overnight; lowering down the PDMS curing temperature (compared to 75 °C) is for avoiding deformation of the PMMA mold since the glass transition temperature for PMMA ranges from 85 to 165 °C.²⁷ Then, the PET film was removed and the array of ring-shaped PDMS structures was de-molded.

In part C, the thin PDMS film (on a silicon wafer) fabricated in part A and the array of ring-shaped PDMS structures fabricated in part B were treated with oxygen plasma (55 W, 5 s) and then the treated parts were brought into contact and baked on a hotplate at 55 °C for 2.5 h to form feasible bonding. Finally, the bonded structures were peeled-off from the silicon wafer and diced into separate valve modules using biopsy punches (4 mm). Multiple valve modules with an inner diameter of 1.5 mm, an outer diameter of 4 mm, and a film thickness of ∼100 μm can be fabricated in a single batch as shown in Figs. 2(c) and 2(d).

B. Design and fabrication of the valve-testing device

The operation characterization of the Luer-lock valves starts with the design of a microfluidic device capable of evaluating the performance of the valves. For convenience, this device will be referred to as the “valve-testing device.” As shown schematically in Fig. 3(a), the valve-testing device consists of two microchannels (section: 300 × 300 μm²), which combine at a Y-shaped junction; each microchannel has a valve port for the configuration of Luer-lock valves. The inlet and outlet of the channels have as-printed barbed connectors for easy tubing,⁹ the detailed design parameters for the barbed connectors can be referenced in previous publications.^10,28,29 The CAD model of the valve-testing device is as shown in Fig. 3(b). The photograph of the device fabricated using a commercial stereolithography machine (Perfactory, EnvisionTEC GMBH) is as shown in Fig. 3(c).

FIG. 3.

Design, fabrication, and characterization of the valve-testing device. (a) Schematically, the valve-testing device consists of two microchannels that combine at a Y-shaped junction; each microchannel has a valve port for the configuration of Luer-lock valves. (b) The CAD model of the valve-testing device. (c) The photograph of the device fabricated using a commercial stereolithography machine. (d) An example of the valve configuration for testing a single valve. (e)–(g) The design and actual images of the inspection window for the microscopic investigation on the fluidic patterns. (h)–(k) The design of the bottom surface morphology of the valve port and fabrication results including (i) an optical- and laser-combined image and (j) the bottom surface plotted using the heat map to denote the depths; (k) The profile plot of the surface height of two vertical cross sections. Scale bars: (c) and (d) 10 mm; (f) 500 μm; (g) 200 μm; (i) and (j) 400 μm.

Several considerations are implemented in the valve-testing device for the test of the Luer-lock valves.

First, the two valve ports of the valve-testing device can be flexibly configured for different purposes; e.g., as shown in Fig. 3(d), for testing a single valve, one valve port can be capped using a male-type Luer cap (XX-WS01K*, Terumo Corporation), while another valve port can be configured as a Luer-lock valve with a valve module and a Luer-to-barbed fitting (VRM106, ISIS Co., Ltd.). During the tests, the inlets of the two microchannels (prefilled with solutions with or without dyes) will be driven under the same constant pneumatic pressure supplied separately from a multichannel pressure controller (MFCS™, Fluigent). In the case of single valve testing [as configured in Fig. 3(d)], if a low control pressure P_C is applied to the Leur-lock valve, the two microchannels are both open and the width of the two combining laminar streams will be the same at the downstream of the fluidic junction; then, by gradually increasing the control pressure P_C applied to the Luer-lock valve, the microchannel with the valve will show gradually increased fluidic resistance and finally be closed under a certain threshold pressure P_threshold.

Second, since the materials of our stereolithography 3D printer are not fully transparent (some are translucent and some are even considered opaque at macroscale), we modified our designs of the microchannels at a certain location as highlighted in Figs. 3(a) and 3(e) and named it the “inspection window,” for the microscopic inspection on the fluidic pattern in the microchannels. Specifically, at these inspection windows, the upper and lower wall thicknesses of the microchannels are thinned to 300 μm to allow more light penetration while maintaining the mechanical strength of the microchannels. Practically, we can investigate the microchannel under bright field using an inverted microscope (IX-71, Olympus) with the default illumination setup, as shown in Figs. 3(f) and 3(g). In addition, the inspection window is also compatible with a high-speed microscope (VW-9000, Keyence Corp.) to allow the high-speed imaging of the laminar flow without significant compromises on the shutter speed setting of the microscope camera.

Third, the geometry of the microchannels that will be closed by the valve module is defined by the bottom surface morphology of the valve port. As shown in Fig. 3(h), the bottom surface has a cone-shaped design to comply with the dome-shaped deformation of the PDMS membrane. The footprint of the Luer-lock valve is decided by the upper edge diameter of the cone shape, which is fixed to 0.8 mm considering both the resolution of the stereolithography machine and the inner diameter of the valve module; since the valve modules are manually punched using biopsy punch, the center of the PDMS membrane (i.e., the center of the inner cylinder of the ring-shaped structure) could be offset from the center of the punch peripheral; thus, the upper edge diameter of the cone shape (0.8 mm) is smaller than the inner diameter of the valve module (1.5 mm) to account for alignment errors during the placement of the valve module. Such a compromise in the valve footprint will result in slow switching response of the valve, which shall be attenuated if automated punching machines with precise position alignment were used for the dicing of the valve modules. The lower circular edge of the cone shape is 190 μm in diameter and the height of the cone shape is 200 μm. The bottom surface morphology of the Luer-lock inlet is scanned using a laser microscope (VK-X, Keyence Corp.) with an optical- and laser-combined image as shown in Fig. 3(i); the bottom surface is plotted using the heat map to denote the depths as shown in Fig. 3(j), and the profile plot of the surface height of the two vertical cross sections [as indicated in Fig. 3(j)] are shown in Fig. 3(k).

C. Operation characterization of the Luer-lock valves

With the valving-test device, the following operational tests were performed to characterize the Luer-lock valves: a static cycling test to evaluate the repeatability of the valves, a dynamic sweeping test to probe the responsiveness of the valves, and a two-fluid switching test to probe the sharpness of switching. Experimental setups were slightly different for these tests: for the static cycling test and the dynamic sweeping test, the valving-test device was configured with one valve port capped using a male-type Luer cap and another valve as a Luer-lock valve, as shown in Fig. 3(d), and the two microchannels, one without valve and prefilled with de-ionized water and another with a valve and prefilled with blue-ink, were driven under a same constant pneumatic pressure (10 mbar); for the two-fluid switching test, the valving-test device was configured with both its valve port as Luer-lock valves, and the two microchannels, driven under a same constant pneumatic pressure (10 mbar), were prefilled with red-ink or blue-ink, respectively. Detailed results of the Luer-lock valve operation characterization are as follows.

For the static cycling test, the control pressures P_C were applied to the valve in a step-by-step fashion, forward and backward between 0 mbar and 150 mbar, with an interval of 10 mbar. The laminar flow pattern at the fluid junction was monitored through the inspection window using a high-speed microscope (VW-9000, Keyence Corp.). The control pressures were kept constant in each testing step and microscopic images were taken after the laminar flow was adequately stabilized, which is both checked by eyes and monitored using the MotionGraph function provided by the high-speed microscope. The MotionGraph shows the sum of relative luminance values of each pixel, which is calculated based on the difference/correlation in brightness or color difference between the image of each frame and a “representative image”; the representative image is automatically selected based on specific algorithms to represent each cycle of the captured periodic motion, as detailed in Ref. 30. The stabilization of the laminar flow, as imaged by the high-speed camera, will result in a plateauing of the MotionGraph. Width fractions of the laminar flow of the blue-ink were calculated using the macro programming function of ImageJ: in detail, first, the background of each image was subtracted with a reference image taken when the valve is fully closed, which resulted in images with only the fluid bands of the blue-ink; then, the net intensity sums of the fluid bands were measured and the width fractions of the laminar flow were calculated by dividing the net intensity sum of the bands. The cycling tests were performed with 20 cycles, and the flow width fractions of the blue laminar flow (i.e., the blue-ink), normalized against the flow width of the blue laminar when the valve was fully open, are plotted against the control pressure as shown in Fig. 4(a). For simplicity, data were plotted with the selected cycles (1st, 5th, 10th, 15th, and 20th); left arrows or right arrows in the legends indicate whether descendent or ascendant control pressures were applied to the Luer-lock valve, respectively. Representative images when the valve was fully open or fully closed are as shown in Figs. 4(b) and 4(c), respectively.

FIG. 4.

Results on the operation characterization of the Luer-lock valves. (a) The cycling tests: the flow width fractions of the blue laminar flow (i.e., the blue-ink), normalized against the flow width of the blue laminar when the valve was fully open, plotted against the control pressure. 20 cycles were performed, and for simplicity, data were plotted selectively (1st, 5th, 10th, 15th, and 20th); left arrows or right arrows in the legends indicate the descendent or ascendant control pressures, respectively. (b) Image of the fluid pattern when the valve was fully open. (c) Image of the fluid pattern when the valve was fully closed. (d) The dynamic sweeping test: sum of relative luminance values of each pixel plotted against time to indicate the response of the Luer-lock valve under alternating control pressures (square waves) with various frequencies. (e) Two-fluid switching test: a long tail of blue-ink thread surrounded by the red-ink was speculated and disappeared up until 80 ms after the switching was triggered. Scale bars: (b) and (c) 400 μm. *Calculated based on the difference/correlation in brightness or color difference between the image of each frame and a “representative image,” which is automatically selected based on specific algorithms to represent each cycle of the captured periodic motion, as detailed in Ref. 31.

For the dynamic sweeping test, control pressure sequences were in the form of square waves (high/low pressure: 150/0 mbar, duty cycle: 50%) with frequencies of 0.5, 1, 2, 3.3, and 5 Hz. The MotionGraphs (as explained in previous texts) of the laminar flow pattern are plotted as shown in Fig. 4(d). Since the transparent flow will result in brighter images compared to the blue-ink flow, wider transparent flow generates higher value in the MotionGraphs, and vice versa. Also, the representative image,³⁰ which is used to generate the curves, is the image frame that has the lowest sum of luminance value in the whole video. These results can be interpreted as follows. First, the threshold of operation frequency for the Luer-lock valve is 1 Hz; for a frequency no higher than 1 Hz, the curves show plateaus (with close-to-equal values) indicating the stabilizing of fluid pattern in each of the wave cycle, while for a frequency higher than 2 Hz, overshooting of the curve starts to emerge since the valves cannot be closed fast enough. Second, the 0.5 Hz curve shows a slightly increased value after the valve was fully open, indicating an increase of the width of transparent flow. This phenomenon can be caused by the gradual inflation of the valve membrane that dissipates the pressure to drive the flow; the phenomenon disappears for the 1 Hz curve since the time for the membrane inflation to develop is not enough. Therefore, these results also suggest that it is better to set a non-zero low control pressure to counter the inflation of the valve membrane. Third, the high plateau values (∼2 × 10⁶) of the curves with frequencies of 0.5 and 1 Hz can be used as a reference to judge how much the valves are closed for the frequencies of 2, 3.3, and 5 Hz: the peak values of the curves do not reach 2 × 10⁶, indicating that the valves are not fully closed during the period of these sweeping tests; also, the peak value decreases as the frequency increases, indicating that the amplitude of membrane deformation (toward the lower channel) decreases as the frequency increases; hence, the valves are less closed with higher operating frequencies.

For the two-fluid switching test, the control pressures applied to the pair of Luer-lock valves were switched between the two valves with a high/low pressure of 140/0 mbar as schematically indicated in Fig. 4(e). As a result, a tail of blue-ink thread surrounded by the red-ink was speculated and disappeared up until 80 ms after the switching was triggered.

In summary, the repeatability of the Luer-lock valve is at an acceptable level as shown with the static cycling test, and the result of the dynamic sweeping test is in agreement with the general knowledge that the responses of the pneumatic valves are ≥1 Hz;³¹ though since the switching test results indicate a theoretical best response on the order of less than 100 ms, there will be an expectable performance boost by further downsizing the footprint of our valve modules using 3D printers with higher resolution and by optimizing the valve design to avoid PDMS membrane inflation caused by the flow underneath the membrane (such as adopting the “doormat” structure).³² In comparison with other types of valves such as piezo-, magnet-, or thermal-actuated valves, our pneumatic type valve has a smaller footprint to allow on-chip integration, and the electrode-free design with easy assembly method can reduce the overall manufacture time and material cost.

D. Formation of encoded core–shell fiber using the Luer-lock valves

To demonstrate the applicability of the Luer-lock valves, a core–shell fiber encoding device is designed with the aim to enable microfluidic automation in the spinning of core–shell microfibers. Recently, microfluidic spinning has emerged as an advanced method to fabricate the cell-laden fibers with diverse shapes and sizes.^33,34 Among these microfluidic methods, the fabrication of “core–shell” microfibers using the double coaxial microfluidic device are especially attractive, since they allow the encapsulation of soft materials such as supra-molecules²⁹ and extracellular matrices (ECMs) with cells.³⁵ However, current methods could only fabricate homogeneous “core–shell” fiber with a single type of core–material composition; though there are methods³⁶ that can serially program “simple” (i.e., with a uniform fiber cross section) microfibers, to apply these methods to fabricate core–shell fibers are challenging since cylindrical channels, which can be easily 3D printed, are preferred for the fabrication of core–shell to avoid clogging during fiber fabrication.^11,37,38

Here, we designed the core–shell fiber encoding device to enable the serial encoding of the core–material composition of the core–shell microfibers, as shown schematically in Fig. 5(a). The device is based on the design of the valving-test device for switching two combining fluidic streams using two Luer-lock valves while changing the outlet to a core–shell fiber spinning device consisting of a double coaxial channel, as shown in Fig. 5(b). To effectively eliminate unwanted mixing of the switching streams, a compact style of design is adopted aiming to minimize the distance from the spot where the two streams combine to the spot where fiber starts to cross-link, and such a design has been used to fabricate core–shell cell-laden microfibers:³⁸ in detail, first, the connection of the outlet and the inlet can be shortened to less than several millimeters since the inlet of the core–shell fiber spinning device and the outlet of the valving-test device can be directly printed to be connected; second, coaxial nozzles of the core–shell fiber device are designed to be cone-shaped and tightly cascaded to eliminate length; third, to enable axial-uniform laminar flow formation in the core–shell fiber device, the sheath streams are delivered in a quadro-symmetric fashion, i.e., each sheath stream splits at two stages to end up in four sub-streams. To facilitate the visibility of the channels, we printed the device using a stereolithography machine (J028, DWS) with a transparent resin (DL370, DigitalWax®, DWS); the photo of a device filled with inks is as shown in Fig. 5(c).

FIG. 5.

The spinning of core–shell fibers with alternative fiber cores enabled by Luer-lock valves. (a) The scheme for the core–shell fiber encoding device to enable the serial encoding of the core–material composition of the core–shell microfibers. (b) CAD design of the device. (c) Photo of a device printed with transparent resin and filled with ink. (d) The manual control of the on/off switching of the core fluids (one dyed flow with blue inks and red fluorescent particles, another flow without any dye) during fiber spinning. (e) The tail of a gradually thinning core fiber at the boundary between switched fiber cores. Inlet: the highlighted fluorescent image with high magnification. (f) The fabricated microfiber with alternating core (pitch length ∼8 mm) with 1 Hz switching frequency. Scale bars: (e) 500 μm (inset: 200 μm); (f) 10 mm.

Using the device, we demonstrate the programming of “core–shell” fibers. The two core fluids, dyed differently, consist of low-viscosity sodium alginate aqueous solutions (1 wt. %, IL-6G, KIMICA Corp.) to minimize the fluid pressure in the microfluidic device; high fluid pressure will require a high control pressure for the valve to close the channel, while the pressure controller (Fluigent, MFCS) used in these experiments has a limited driving capability (1000 mbar per channel for four channels). For the formation of alginate fibers, first, the flow of calcium chloride solution (100 mM) and the shell solution (2 wt. % sodium alginate, IL-6G, KIMICA Corp.) are driven by a multi-channel syringe pump under the flow rate of 800 μl/min and 32 μl/min, respectively; then, the two core fluids will be driven using the pressure controller (same as the one for driving the valves) under the constant pressure of 100 mbar. Detailed and basic protocols for spinning core–shell microfibers can be found in our previous publications.^35,38 For the Leur-lock valves, the high/low control pressure to turn off/on the flow of the channel was set to 250/10 mbar; a non-zero low pressure was set to avoid the inflation of the PDMS membrane of the valve caused by the fluidic pressure beneath the membrane; the inflation of membrane will cause over-shooting during the switching since the additional amount of liquid will be stored in the inflated valve since the membrane can be deformed by the fluid beneath and the operation characteristic of the pump has to be fine-tuned according to the flow rate and viscosity. Also, notice that the pressure settings involved in the spinning of the microfibers are higher than the one for the operation characterization of the Luer-lock valves since the viscosity of the sodium alginate solution is 50–60 times higher than of water. Although this fluidic pressure will also change the resonant characteristic of the membrane, its influence can be avoided by operating the switching at a frequency away from the resonant frequency of the membranes.

The fabrication results of the core–shell microfibers are as follows.

First, we confirmed that the switching of the fiber can be achieved by applying control pressure manually. The valve ports of the device were configured to be the same for the test of a single valve: one valve port was capped using a male-type Luer cap and another valve was configured as a Luer-lock valve. The barbed inlet of the valve was connected to a syringe (filled with air) so that the control pressure can be applied manually by pressing the syringe. As a result, we confirmed the manual control of the on/off switching of the core fluids (one “dyed flow” with blue inks and red fluorescent particles, another flow without any dye); as shown in the non-fluorescent images in Fig. 5(d), the channel for the dyed flow, which is controlled by the valve and is initially open, was first closed (which resulted in the stop of the blue thread) by pressing the syringe and then opened (which resulted in the re-start of the blue thread) by releasing the syringe. We investigated the fabricated fibers and found a tail of a gradually thinning core the fiber in the microscopic images, as shown in Fig. 5(e). Such tails resulted from the lagging of the switching and were predictable according to the results of the two-liquid switching tests discussed in Sec. III C.

Second, we performed the fabrication of “core–shell” fibers with alternative core composition under computer-aided valve control (Labview, NI; Fluigent, MFCS, Fluigent) and show that the pitch length of the alternative cores can be altered by tuning the switching frequencies (i.e., 0.5 Hz and 0.1 Hz), as shown in Mov. S1 (0.5 Hz) and Mov. S2 (0.1 Hz) in the supplementary material. In addition, to further facilitate the visualization, the core without dyes were loaded with microbeads (200 nm) to render a milky color in contrast with the other core fluid dyed with blue-ink; the fabricated microfiber with alternating core (pitch length ∼8 mm) with 1 Hz switching frequency is shown in Fig. 5(f). Since previous results [Fig. 4(d)] indicated that the threshold of operation frequency for the Luer-lock valve is 1 Hz, switching frequencies of higher than 2 Hz were not tested for the fabrication of the microfibers. Though a smaller pitch length of the alternative fiber core was not achieved due to the relatively slow response of the current version of the valve, as discussed before, we believe that a smaller valve footprint and improved valve design can further decrease the pitch length of the fiber cores. Despite the fact that many fluid supply systems can enable switching between one line and another to achieve encoding upstream in the fluid supply,³⁹ the response of control will be much slower compared to the on-chip valves; the inertial of the pressure built-up in the fluid stream will result in slower switching and a longer tail between the switching streams. Therefore, we believe our on-chip valves can enable faster switching of the fibers on-a-fly, which could lay the foundation of the fiber-based hydrogel 3D printers with drastically increased resolution compared to the current multi-material 3D printers.^6,39

IV. CONCLUSION

In this work, we proposed a type of “Luer-lock” valve—an elastomeric valve that is easy-to-configure in 3D. The valves were prefabricated using PDMS molding and can be embedded into 3D printed microfluidic devices by screwing tight the Luer-lock connectors. In the experiments, we validated the design and functionality of these valves (footprint diameter: 0.8 mm) with a shut-off pressure of 140 mbar and a maximal switching frequency of ∼1 Hz. We demonstrated that the Luer-lock valves can be integrated into a 3D printed double coaxial microfluidic device to enable the automatic fluid switching functions and fabricated serially encoded core–shell hydrogel microfibers with a switching fiber core (pitch length ∼8 mm), which are potentially useful for biofabrication, tissue engineering, and regenerative medicine purposes. Future works will focus on improving the response of our valve by further downsizing the footprint of our valve modules using stereolithography printers with higher resolution and optimizing the valve design to avoid PDMS membrane inflation caused by the flow underneath the membrane. Also, since the Luer-to-barbed fitting can connect not only gas but also electrical wires and optical fibers, our Luer-lock based approach can potentially be used for integrating optical fiber-based detectors or electrodes into 3D printed microfluidic devices.

SUPPLEMENTARY MATERIAL

See the supplementary material for videos Mov. S1 and Mov. S2 demonstrating the switching the of fiber cores with switching frequencies, i.e., 0.5 Hz and 0.1 Hz, respectively.

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author upon reasonable request.