Academic Article

Rapid Fabrication by Digital Light Processing 3D Printing of a SlipChip with Movable Ports for Local Delivery to Ex Vivo Organ Cultures

by

Rapid Fabrication by Digital Light Processing 3D Printing of a SlipChip with Movable Ports for Local Delivery to Ex Vivo Organ Cultures

Megan A Catterton, Alexander G Ball, and Rebecca R Pompano

SlipChips are two-part microfluidic devices that can be reconfigured to change fluidic pathways for a wide range of functions, including tissue stimulation. Currently, fabrication of these devices at the prototype stage requires a skilled microfluidic technician, e.g., for wet etching or alignment steps. In most cases, SlipChip functionality requires an optically clear, smooth, and flat surface that is fluorophilic and hydrophobic. Here, we tested digital light processing (DLP) 3D printing, which is rapid, reproducible, and easily shared, as a solution for fabrication of SlipChips at the prototype stage. As a case study, we sought to fabricate a SlipChip intended for local delivery to live tissue slices through a movable microfluidic port. The device was comprised of two multi-layer components: an enclosed channel with a delivery port and a culture chamber for tissue slices with a permeable support. Once the design was optimized, we demonstrated its function by locally delivering a chemical probe to slices of hydrogel and to living tissue with up to 120 µm spatial resolution. By establishing the design principles for 3D printing of SlipChip devices, this work will enhance the ability to rapidly prototype such devices at mid-scale levels of production.

Keywords: SLA printing, resin printing, tissue culture, local stimulation, two-phase microfluidics

We kindly thank the researchers at University of Virginia for this collaboration, and for sharing the results obtained with their system.

Introduction

The ability to produce microchips easily and with minimal manual assembly, while retaining rapid prototyping capabilities, is highly desirable for pushing microfluidic devices past the first hand-built prototype stage [1,2,3]. Scaled-up fabrication is critical to conducting experiments at moderate scale (dozens of devices) and for propagating such technology to collaborators. In particular, this scale of fabrication would be useful for SlipChips, which are two-phase, reconfigurable microfluidic devices [4,5,6,7,8,9]. SlipChips usually comprise two planar components that can be “slipped” relative to one another, contain recessed features to hold droplets or streams of aqueous solution, and are separated by a thin layer of oil [4]. SlipChip devices were first developed in the Ismagilov lab [4] as a new technology to perform in low-resource settings [5,6,7]. The first SlipChips were fabricated from glass plates, which offer ideal surface properties and optical clarity but require wet etching with HF, a hazardous procedure that requires a skilled technician [4,10]. Since then, many different Slip-based designs have evolved, including rotational Slipdisc and paper-based SlipPADs, to perform a wide range of laboratory processes such as PCR, cell culture and local delivery to tissue slices [8,9,11,12,13,14,15,16,17]. Fabrication is especially challenging for novel slip-based devices that have multiple layers per component [9,17]. Although injection molding can simplify fabrication at large scale [18], an alternative method is needed to fabricate SlipChips at a moderate scale, while retaining the ability to rapidly prototype.

Any fabrication system for SlipChips must be able to meet four platform requirements, in addition to producing the specific features needed for the intended application. To prevent the aqueous phase from spreading into the oil-filled gap between components, high capillary pressure at the oil–water interface must be maintained. Therefore, the surfaces in contact with the oil layer must be flat and smooth enough to create a gap height of ~1–10 µm across the entire face of the chip [5]. Furthermore, these surfaces must be hydrophobic; if a fluorinated oil is used [4], then a fluorophilic surface is preferred. Finally, for SlipChips that rely on visual alignment or optical detection, the layers must be optically transparent.

Considering these requirements, we reasoned that digital light projection (DLP) 3D printing, which uses UV or blue light to cure photocrosslinkable resins layer by layer [19,20], may facilitate SlipChip fabrication and allow for rapid prototyping. This additive method is quickly gaining popularity for fabricating small parts and microfluidic devices, because of both its high feature resolution and reproducibility and its rapid fabrication speed compared to traditional soft-lithography [3,21,22,23]. While 3D printing has not been reported previously for SlipChips, two of the four fabrication requirements are already met. We recently described a method for fluorination of a DLP-printed surface based on solvent-based deposition of a fluoroalkyl silane [24], and others have demonstrated optically transparent parts by printing clear resins on a glass surface to reduce light scattering [25].

As a case study for fabrication of a SlipChip by 3D resin printing, we considered a microfluidic movable port device (MP device) previously developed by our lab for local stimulation of ex vivo organ slices at user-selected locations [9]. The MP device is a SlipChip that is comprised of two multilayer components: a bottom component containing a simple enclosed microchannel that terminates in a single, vertical delivery port (delivery component), and a top component featuring a semipermeable tissue culture well (chamber component) (Figure 1a). A bolus of aqueous solution is pumped into a specific region of a tissue slice by aligning the delivery port to a port in the culture well (Figure 1b). Local delivery devices like this one have been used to study intrinsic tissue properties and to screen for potential drugs [9,26,27,28,29,30]. Compared to a device with stationary ports, the SlipChip functionality of the MP device lessens the amount of user handling of a tissue slice and allows more flexible on-demand selection of the delivery region. However, in the original hand-built prototype, an extensive fabrication process limited the accessibility and distribution of the MP device to other labs and collaborators [9].

Here, we established an approach to fabricate a 3D printed SlipChip for the first time, using the MP device as a case study. First, we validated the selection of a DLP resin designed for microfluidic devices to meet the optical transparency, surface roughness, surface chemistry, and biocompatibility requirements of the tissue-specific movable port device. Next, the device design was optimized to maximize the functionality of the required ports and channels while minimizing the fabrication time complexity with DLP printing. The ability of the assembled device to deliver aqueous solutions without leaks into the gap was tested, and finally, we tested the ability to stimulate live organ cultures locally and with the position selected on demand.

2. Materials and Methods

2.1. Device Design, 3D Printing, and Laser Etching

All 3D printed parts were designed using Autodesk Inventor 2018 (Mill Valley, CA, USA). The CAD files (in Supporting Information) were sliced at 50 µm intervals using MII Utility Shortcut V 3.27 and printed using a CADworks3D M50-405 printer (30 µm xy-resolution, CADworks3D, Toronto, ON, Canada) in BV-007A resin (MiiCraft, via CADworks 3D). The printer setting for the BV-007A resin at a 50 µm slice height was a slow peeling speed, 0.1 mm gap adjustment (unless printing on glass which was 0.27 mm), 1.15 s curing time, 1 base layer, 9.0 s base curing time, 1 buffer layer, and 75% light power. To print parts on glass, a cover glass slide, 36 mm × 60 mm with a thickness of 0.13–0.17 mm (Ted Pella, Redding, CA, USA), was attached to the baseplate by curing a thin layer of BV-007A with a 405 nm UV light (Amazon, Seattle, WA, USA) [25]. The parts were rinsed with methanol (Fisher Chemical, Waltham, MA, USA) and post-cured in a UV light box for 20 s. No additional leaching steps were applied to the printed pieces used in this work. In preliminary experiments, we found that solvent washes at varied temperatures or extended UV light exposure did not substantially improve the biocompatibility of the BV-007A resin. To complete the chamber component, an array of ports with an 80 μm diameter were laser etched (Versa Laser 3.5, Universal Laser Systems, Scottsdale, AZ, USA) into the printed BV-007A part, using a power setting of 7% and a speed of 10%.

2.2. Fluorination of Resin Surface and Contact Angle Measurements

Parts printed in BV-007A were silanized using our recently described method [24]. The parts were submerged into a 10% v/v solution of tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane (Gelest Inc., Morrisville PA, USA) in Fluorinert FC-40 (Sigma Aldrich, St. Louis, MO, USA) for 30 min at room temperature. The surfaces were rinsed with 95% ethanol (Koptec) and DI water and finally dried with a nitrogen gun.

Surface air–water contact angles and three-phase contact angles were measured on cubic printed pieces (8 × 8 × 15 mm3) using a ramé-hart goniometer (model 200-00, ramé-hart instrument co., Succasunna NJ, USA) and DROPimage Advanced software (ramé-hart instrument co., Succasunna, NJ, USA). For consistency, the smooth, flat face of the cube produced against the polytetrafluoroethylene (PTFE) sheet was tested in all cases; this was also the side of the print that faced the oil layer in the SlipChip. The contact angle was measured in triplicate (3 separate printed pieces per condition), by pipetting one 5 µL droplet of 1× phosphate buffered saline (PBS) (Lonza, Walkersville, MD, USA.; DPBS without calcium or magnesium) onto the printed surface. For three-phase contact angle, the printed cube with a droplet was inverted into a cuvette filled with FC-40 oil containing 0.5 mg/mL triethyleneglycol mono[1H,1H-perfluorooctyl]ether (RfOEG). RfOEG was synthesized in house as reported previously (see Supporting Methods) [9,31,32].

2.3. Surface Profilometry

To assess surface roughness, the root mean square deviation of the surface height of the printed parts was measured with a Zygo optical surface profilometer (Zygo, Berwyn, PA, USA) at the Nanoscale Materials Characterization Facility at the University of Virginia. Cubes of 8 × 8 × 8 mm3 were printed, and surface roughness was measured on all sides, specifically the surfaces printed against the aluminum baseplate or printed against glass, closest to the PTFE sheet at the bottom of the vat, and the sides of the print. As a positive control, a glass microscope slide was also analyzed after plating with 30 nm of Au/Pd by a Technics sputter coater (Technics).

2.4. Measurement of Curvature of Printed Pieces

Images of the side profiles of 3D printed 30 × 30 mm2 prisms of varied height (2–5 mm) were collected using a Zeiss AxioZoom microscope (Jena, Germany). The displacement from horizontal due to curvature was manually measured in Zen 2 software (Zeiss, Jena, Germany).

2.5. Animal Work and Tissue Slice Collection

All animal work was approved by the Institutional Animal Care and Use Committee at the University of Virginia under protocol #4042, and was conducted in compliance with guidelines of the Office of Laboratory Animal Welfare at the National Institutes of Health (United States). Both male and female C57BL/6 mice aged 19–21 weeks (Jackson Laboratory, Bar Harbor, ME, USA) were housed in a vivarium and given water and food ad libitum. Lymph nodes were harvested from the mice following humane isoflurane anesthesia and cervical dislocation. The tissues were sliced according to a previously published protocol [33]. Briefly, peripheral lymph nodes were collected and embedded in 6% w/v low melting point agarose (Lonza, Walkersville, MD, USA) in 1× PBS. After the agarose had hardened, agarose blocks containing lymph nodes were extracted with a 10 mm tissue punch (World Precision Instruments, Sarasota, FL, USA). The blocks were mounted with super glue on a stage and sliced into 300 μm thick sections using a Leica VT1000S vibratome (Bannockburn, IL, USA) in ice-cold 1× PBS. The lymph nodes were sliced at a speed setting of 90 (0.17 mm/s) and frequency of 3 (30 Hz). Slices were cultured in “complete RPMI”: RPMI 1640 (Lonza, 16-167F) supplemented with 10% FBS (VWR, Seradigm USDA approved, 6 89510-186), 1× L-glutamine (Gibco Life Technologies, 25030-081, Waltham, MA, USA), 50 U/mL Pen/Strep (Gibco), 50 μM beta-mercaptoethanol (Gibco, 21985-023), 1 mM sodium pyruvate (Hyclone, GE USA), 1× non-essential amino acids (Hyclone, SH30598.01), and 20 mM HEPES (VWR, 97064–362). Slices of 6% agarose were collected in a similar manner but were stored in 1×PBS instead of complete media.

2.6. Analysis of Tissue Viability

To assess the viability of lymphoid tissue slices after a brief exposure to BV-007A, tissue slices were incubated in 1× PBS in a 3D printed culture well (30 mm × 30 mm × 5 mm printed part, with a central 10 mm-diameter well) for 15 min at room temperature. Then, the slices were moved from the printed substrate into a 24-well plate (VWR) and cultured in “complete media” for 4 h at 37 °C with 5% CO2 to allow time for any delayed effects of on-chip exposure, such as toxicity mediated by protein transcription or translation, to occur. Following a previously established protocol [33,34], the viability of live lymph node tissue slices was assessed by flow cytometry. Briefly, individual slices were crushed to generate cell suspensions. Cells were stained with 75 μL of 67 nM Calcein AM (eBioscience, San Diego, CA, USA) in 1× PBS for 20 min at 37 °C. Stained samples were washed by centrifugation at 400 g for 5 min and resuspended in 1× PBS + 2% FBS (flow buffer). 7-AAD (AAT Bioquest, Sunnyvale, CA, USA, 5 μg/mL final concentration) was then added to the cell suspension. The samples were run on a Guava easyCyte 4-color cytometer (EMD Millipore, 6-2L, Burlington, MA, USA) and analyzed using Guava® InCyte™ Software (EMD Millipore, Burlington, MA, USA). Single stain compensation controls were run on cells from crushed lymph node slices. The Calcein-AM single stain contained a 1:1 mixture of Calcein-labelled and unstained live cells. The 7-AAD single stain contained a 1:1 mixture of live and killed cells; the latter were prepared by treating cells with 35% ethanol for 10 min. Calcein positive and 7-AAD negative cells were defined as viable cells.

2.7. Assembly and Local Delivery with the 3D Printed Slipchip

Prior to assembling the SlipChip, the channel in the delivery component was filled using pressure-driven flow via a Chemyx syringe pump (Fusion 200, Houston, TX, USA). A 0.5 mg/mL solution of FITC-conjugated dextran (150 kDa and 70 kDa for agarose and tissue deliveries experiments, respectively) was flowed into the channel using a 50 μL Hamilton syringe (model 1705 RN; 26 s gauge, large hub needle) and non-shrinkable PTFE TT-30 tubing (0.012” I.D., 0.009” wall thickness, Weico Wire, Edgewood, NY, USA). Next, 500 µL of FC-40 oil containing 0.5 mg/mL RfOEG was pipetted onto the top face of the filled delivery component. The chamber component was lowered onto the delivery component, and the two components were clamped together with two binder clips, sandwiching a thin layer of oil between them. The culture chamber on the top of the chip was then filled with 1× PBS. A sample of agarose gel or tissue was placed into the chamber and weighed down using a small stainless-steel washer (10 mm O.D. and 5.3 mm I.D., Grainger, Lake Forest, IL, USA). The chamber component was manually slipped relative to the delivery component and visually aligned under a microscope to align to a desired port. To initiate a delivery, the syringe pump was turned on at the desired flow rate. After 5 s, the pump was turned off and the device was slipped away, to reposition for another delivery or to a reach a closed position. After all deliveries were complete, the sample was removed, and the chamber was flushed with 1× PBS and refilled for the next sample. All delivery experiments were performed at room temperature.

All deliveries were monitored in real time using a Zeiss AxioZoom upright microscope with a PlanNeoFluor Z 1×/0.25 FWD 56 mm objective, Axiocam 506 mono camera and HXP 200 C metal halide lamp (Zeiss, Jena, Germany), using filter cubes for GFP (Zeiss filter set #38), and Violet Chroma Filter (49021, ET-EBFP2). Images (16 bit) were collected before, during, and after delivery. During deliveries, time lapse images were collected at 1 s intervals. All images were analyzed in Zen 2 software (Zeiss, Jena, Germany).

2.8. Analysis of Delivery Widths

After alignment of the delivery port to an array port, a 5 s pulse of fluorescein (FITC)-labeled 150 kDa dextran was delivered to a 6% agarose slice at flow rates ranging from 0.2 to 1 μL min−1 (n = 3). After delivery, the device was slipped prior to imaging, to avoid the fluorescent signal from the underlying channel. Delivery width was determined from image analysis as previously described [26]. Briefly, line scans were drawn radially across the delivery region, and the background autofluorescence of the resin was subtracted. The data were fit to a Gaussian curve in GraphPad Prism version 8 (San Diego, CA, USA). The width was defined as two standard deviations of the Gaussian curve.

To fit the curve of the spread of analyte with respect to time, we used a previously published analytical model [9]. First, we assumed that the volume delivered per unit time was described by a cylinder:

where w [μm] is the width (diameter) of the delivery, h [μm] is the height of the slice, Q [μL/min] is the volumetric flow rate set by the pump, and Δt [sec] is the length of time of delivery. Solving for width gives Equation (2):

2.9. Delivery to Lymph Node Tissue

The device was assembled and a lymph node slice was placed into the chamber. A 5 s pulse of FITC-labeled 70 kDa dextran was delivered at a flow rate of 0.25 μL min−1. After the first delivery, the device was repositioned and another delivery was performed. This was repeated for four different slices and with slight variations in the number of deliveries on three separate occasions.

Results

3.1. Design Goals for a 3D Printed SlipChip with Movable Ports

The movable port device consisted of two components: a chamber to hold a tissue slice with a porous support in the form of a port array, and a delivery component with an enclosed channel with a small terminating port (Figure 1a). To assemble the device, the microchannel in the delivery component was filled with aqueous solution, and the chamber component was lowered on top while carefully sandwiching a layer of immiscible oil in between. To operate the device, the delivery port was aligned with a port in the array above, and pressure driven flow is used to deliver a short pulse of fluid into the tissue in the chamber (Figure 1b).

Before a MP device could be rapidly fabricated by DLP printing, there were two major challenges to be addressed (Figure 1c). The first challenge, applicable to any 3D printed SlipChip, was to have a small gap height between the printed parts to prevent leaking into the oil layer during delivery. To achieve a small gap height, the two surfaces closest to the oil gap must be both smooth and flat across the width of the component (30 mm) (Figure 1c). Flatness can be challenging because photocurable resins shrink when crosslinked, inducing mechanical stress that warps the print if not addressed in the print design [35]. Furthermore, an array of microscale ports and enclosed microchannel had to be integrated without disrupting the flat surface [36]. The second challenge, specific for biological applications, was biocompatibility of the printed resin with tissue slices housed in the delivery component (Figure 1c). The question of resin toxicity is of great interest to the microfluidics community and is still under active investigation [25,37,38].

3.2. Selection of Materials and Print Conditions for Transparency, Smoothness, Fluorination, and Cytocompatibility

Before designing the microfluidic device, we first selected and validated a resin for its suitability for the intended use in the SlipChip. We chose to use BV-007A resin because of its ability to generate microfluidic devices with high feature resolution [39,40]. First, we addressed the surface roughness and optical transparency of the DLP printed parts. While the polymeric surface would never be as smooth as glass, prior SlipChips have included microposts to set a defined gap height, e.g., of 2 µm, between two glass components [5]. Therefore, we hypothesized that surface roughness ≤ 2 µm would provide an acceptably small gap height. Surface roughness was expected to differ across the various faces of a printed piece, e.g., the bottom that is printed against the baseplate or against glass, the sides of the print, and the top that prints in contact with the PTFE sheet lining the vat (Figure S1a). As expected, optical profilometry showed that the surface printed against the rough aluminum baseplate and the sides of the printed piece were rough, with RMS (root mean square of surface height) > 3 µm (Figure 2a). The polymeric faces printed against glass and PTFE were much smoother, with RMS ≤ 0.3 µm (Figure 2a). For reference, glass itself had a surface roughness of 5 ± 0.5 nm (n = 3). From these data, we concluded that the print for a SlipChip must be oriented such that the surfaces intended to contact the oil gap were printed against glass or the PTFE sheet. Additionally, we also tested for optical transparency, which was desirable for visual alignment of channels and ports in the SlipChip. As previously described [25], printing against glass provided optical transparency, whereas printing against the rough aluminum baseplate yielded an opaque sample (Figure 2b).

Materials

Clear Microfluidics Resin V7.0a

Learn More

H SeriesM

To prevent spreading of aqueous solution between the components in the oil gap, the surface chemistry of the chip must be fluorophilic and hydrophobic where it contacts the oil phase. While the BV-007A resin yields parts that are moderately hydrophilic, we recently described a method for fluoroalkyl silanization for SLA resins [24]. Here, this method was applied to silanize the BV-007A, by placing the surface to be silanized in a solution of 10% fluoroalkyl silane in FC-40 oil (see Methods). We confirmed that silanization not only increased the three-phase contact angle of a 1× PBS droplet resting on the surface in air, but also when immersed in FC-40 oil (Figure 2c). The water/oil/resin contact angle of >115° indicated a highly hydrophobic surface [9].

Finally, as the MP device was intended to be used with live organ slices, we sought to identify conditions in which tissue viability was not affected by the BV-007A printed pieces. Ex vivo slices of murine lymph node tissue were used in these experiments, as we have previously characterized local delivery to such tissues [9,26]. During use of the movable port device, the tissue slice is in contact with the surface of the 3D printed chamber for only a few minutes, typically <5 min for alignment and <10 s for the delivery. In separate work, we have shown that multi-hour physical contact of murine splenocytes with parts printed in BV-007A was cytotoxic after just 4 h [41]. Therefore, we restricted this study to an exposure period of 15 min, which represents triple the expected exposure time for tissue spent in contact with the resin during use of the device. Tissue viability after 15 min exposure was comparable to that of off-chip controls (Figure 2d and Figure S2). As this viability test was based on membrane integrity and esterase activity, further tests for cell and tissue function may be appropriate depending on the intended application of the chip. We and others continue to work to identify a resin or a post-print treatment strategy that provides biocompatibility with primary tissues for longer time periods, while still maintaining with the high print resolution of BV-007A [25,37,38,42].

3.3. Optimizing the Design and Printability of the Delivery Component

Having identified the material and conditions for SlipChip function, we turned to designing the components of the movable port device. The delivery component required three key features to be printed while maintaining a smooth, flat surface: an interior channel, a delivery port, and an inlet (Figure 3a). Additionally, alignment markers (small inset wells on the top of the component) were included in the design to aid in visual alignment of the device when delivering to opaque tissues. Although it is common practice to print at angle to achieve higher resolution for interior channels (Figure S1b, angled) [43,44], the requirement for smoothness dictated that the design be printed horizontal relative to the baseplate, such that the gap-facing surface was printed against PTFE (Figure S1b, flat). In addition, the requirement for a flat profile to minimize gap height meant contending with shrinkage and associated deformation during photocrosslinking [45]. In addition to rounding the sharp corners, we found that increasing the thickness (z) of the printed part was required to minimize mechanical stress during printing for a part with a 30 × 30 mm2 footprint [45,46]. The thickness of the print was varied from 2 to 5 mm, and the delivery component required at least 5 mm thickness to prevent the part from curling (Figure 3b,c).

Having optimized the print geometry and overall dimensions of the piece, the design of the enclosed channels and ports were then optimized to minimize the channel cross-section while retaining printability (Figure 3d). We share these details to aid other researchers who are also working at the limits of the resolution of a 3D printer. To reduce blockage during printing, the channel was positioned close to the top of the part to minimize UV-exposure from subsequent layers, which is a particular issue for transparent resins. Additionally, the length of the channel was minimized (15 mm), because longer channels were more difficult to clear of uncrosslinked resin through the small terminating delivery port. To minimize reagent volume during use, we minimized the cross-sectional area of the channel. In a test piece printed with a series of 15 mm channels of varied cross-sectional size and a square or diamond shape, the minimum cross-section that remained open was 0.5 × 0.5 mm2 in both shapes (Figure 3d and Figure S3). Thus, a 0.5 × 0.5 mm2 cross-section was selected, and the square was selected over the diamond shape in order to minimize the horizontal width the channel during optical imaging of the device. The diameter of the delivery port was optimized in the same test piece, with a series of ports of varied diameter atop each channel (Figure 3d, inset). All ports with diameter ranging from 0.15 to 0.35 mm were successfully printed, with close fidelity (<10% error) to drawn dimension (Figure 3e). The smallest printable port was 0.138 ± 0.009 mm (drawn diameter 0.15 mm); ports drawn smaller failed to print (data not shown). Finally, we designed a simple press-fit female port to ensure a snug fit with the microfluidic tubing at the inlet (0.78 mm OD PTFE tubing), by printing mock inlets of 0.76–0.87 mm drawn diameter (Figure 3f). A 0.80 mm drawn diameter port was determined to give a snug fit with the tubing. In summary, the optimized design for the inlet, enclosed channel, and terminal delivery port (0.8 mm inlet, 0.5 × 0.5 mm2 square channel cross-section, 0.15 mm drawn delivery port) yielded a 3D printed delivery component that could be reproducibly printed and was sufficiently flat and smooth (Figure 3g).

3.4. Optimization to Minimize Port Size and Preserve Optical Transparency of the Chamber Component

The top component of the MP device included a chamber (12 mm diameter) to hold tissue samples in media, with a permeable support at the bottom for delivery of fluid from below. Based on our prior work, the ports in the chamber component needed to be in the range of 0.070–0.110 mm, i.e., large enough to minimize flow resistance and small enough to create a localized delivery [9]. The support needed to be transparent for visual alignment, and the requirement for smoothness meant that the bottom of the chamber component needed to be printed against glass or the Teflon vat.

We originally tested a one-step fabrication method for this component, by embedding a membrane or mesh support into the part during 3D printing or by directly printing the port array (Figure 4a,b). We found it simple to embed a nylon or metal mesh in the component by adhering it to the baseplate or glass prior to printing (Figure 4a and Figure S4). Unfortunately, due to resin shrinkage during polymerization, the mesh did not remain taut, preventing its use in the SlipChip. Next, we attempted to directly print the small ports in an array (Figure 3b), but it was challenging to meet the requirements for both small port size and transparency. Orienting the port array against glass on the baseplate proved unfeasible due to the required overexposure of the first layers of the printing, which lowered the spatial resolution in these layers. On the other hand, orienting the port array as an overhang generated ports with an acceptable diameter (~110 µm), but the unsupported overhang led to stretching and distortion, which reduced transparency (Figure 4b).

Since fabricating the port array in a single step proved challenging, we elected to use a two-step process (Figure 4c). First, the chamber component was 3D printed with the solid bottom of the chamber well (200 µm thick) oriented against the glass. Second, a CO2 laser was used to etch a port array into the bottom of the chamber. The laser-etched ports had a diameter of 0.081 ± 0.002 mm (n = 74), well within the acceptable range, and the entire array was etched in < 1min. Additionally, unlike the accumulation of melted plastic observed when laser etching acrylic [9], there was no deformation of the BV-007A polymer during laser etching on either side of the chamber components (Figure S5), thus minimizing gap height in the SlipChip. The component was sufficiently transparent for visual alignment. Thus, this straightforward fabrication strategy produced a flat, smooth, monolithic part with a well-defined port array, ready for integration into the final SlipChip (Figure 4d,e).

3.5. The Assembled 3D Printed SlipChip Delivers Fluid without Leakage into the Gap

Having fabricated both components, we assembled the 3D printed SlipChip (Figure 5a,b) and tested its ability to perform local deliveries with leakage of aqueous solution into the oil-filled gap, a critical design goal. To test that the aqueous solution did not leak into the oil gap during use, the delivery port was aligned with a port in the array, and a short pulse of fluorescent dextran solution was delivered to an agarose slice through each of three different array ports (Figure 5c). During and after each delivery, fluorescent and brightfield imaging were used to visually inspect the gap area for the appearance of an interface between aqueous and oil phases, which would indicate a leak. No such interface was observed in 3 separate chip assemblies (9 out of 9 deliveries), indicating a robust capillary-pressure-mediated barrier to leakage. This robust interfacial barrier was also not affected by the size difference between the delivery port (0.138 ± 0.009 mm) and the ports in the chamber array (0.081 ± 0.002 mm).

4. Discussion and Conclusions

This paper describes the requirements and fabrication strategy to achieve 3D printed SlipChips for the first time, and demonstrates 3D printing of a SlipChip device with a movable port for local stimulation of organ cultures as a case study. After optimization, DLP 3D printing produced smooth (RMS ≤ 0.3 µm), flat surfaces that were chemically modifiable by fluorination. The resin parts were biocompatible with the short-term (<15 min) exposures needed for local delivery to tissue slices. The delivery component was designed to be printed in a single step and contained an inlet, enclosed channel and small (0.138 ± 0.009 mm) terminating delivery port. The chamber component was produced via 3D printing followed by laser etching, which provided a monolithic culture chamber with an array of 0.081 ± 0.002 mm diameter ports at the bottom, while maintaining a smooth, flat interface for slipping. The device was able to perform multiple slipping and delivery steps without leakage in between the components. The spread of the delivery was dependent on the rate of pump-driven fluid flow, and the resolution was sufficient to target substructures in multiple locations in a tissue slice.

We anticipate that this DLP 3D printing fabrication method will enable the polymeric SlipChips, and in particular the movable port technology, to become accessible to other labs, by greatly simplifying the fabrication steps, materials, and time. Continued advances in biocompatibility of DLP resins [47,48,49] may eventually enable longer-term culture on the 3D printed chip, which was a limitation here. Furthermore, while the current device used binder clips, visual alignment, and manual slipping, 3D printing may enable rapid iteration in the future of other clamping methods and pre-programmed integration with manipulators. Thus, rapid prototyping with DLP 3D printing is expected to accelerate advances in movable port technology as well as other SlipChip device designs.

Selective fluorination of the surface of polymeric materials after stereolithography 3D printing

by

Selective fluorination of the surface of polymeric materials after stereolithography 3D printing

Megan A. Catterton, Alyssa N. Montalbine, and Rebecca R. Pompano

With the microfluidics community embracing 3D resin printing as a rapid fabrication method, controlling surface chemistry has emerged as a new challenge. Fluorination of 3D printed surfaces is highly desirable in many applications due to chemical inertness, low friction coefficients, anti-fouling properties and the potential for selective hydrophobic patterning. Despite sporadic reports, silanization methods have not been optimized for covalent bonding with polymeric resins. As a case study, we tested the silanization of a commercially available (meth)acrylate-based resin (BV-007A) with a fluoroalkyl trichlorosilane. Interestingly, plasma oxidation was unnecessary for silanization of this resin, and indeed was ineffective. Solvent-based deposition in a fluorinated oil (FC-40) generated significantly higher contact angles than deposition in ethanol or gas-phase deposition, yielding hydrophobic surfaces with contact angle > 110° under optimized conditions. Attenuated Total Reflectance-Fourier Transform Infrared (ATR-FTIR) spectroscopy indicated that the increase in contact angle correlated with consumption of a carbonyl moiety, suggesting covalent bonding of the silane without plasma oxidation. Consistent with a covalent bond, the silanization was resistant to mechanical damage and hydrolysis in methanol, and was stable over long-term storage. When tested on a suite of photocrosslinkable resins, this silanization protocol generated highly hydrophobic surfaces (contact angle > 110°) on three resins and moderate hydrophobicity (90 – 100°) on the remainder. Selective patterning of hydrophobic regions in an open 3D-printed microchannel was possible in combination with simple masking techniques. Thus, this facile fluorination strategy is expected to be applicable for resin-printed materials in a variety of contexts including micropatterning and multiphase microfluidics.

Keywords: Two-phase microfluidics, Droplet microfluidics, low surface energy, Digital Light processing (DLP), stereolithography printing (SLA)

We kindly thank the researchers at University of Virginia for this collaboration, and for sharing the results obtained with their system.

Introduction

The microfluidics community has increasingly adopted 3D printing for device fabrication, including with fused deposition modeling1–3 and with resin-based methods such as stereolithography (SLA) and digital light processing (DLP) printing.4,5 As a result, methods to control the surface chemistry of 3D printed devices are emerging as a critical challenge, especially for microscale features produced by resin printing.6 In resin printing, UV/visible light is used to cross-link a photocurable, polymeric resin in a layer-by-layer fashion to produce a 3D structure.4,5 While methods for surface functionalization are well established for traditional materials such as glass and polydimethylsiloxane (PDMS), those methods do not necessarily translate directly to the polymeric materials used for 3D printing. A particular challenge is to generate a fluorinated surface on a 3D printed chip. Fluorinated surfaces offer many advantages for microfluidic device design, such as controlled surface wettability for passive fluidic control, chemical inertness, resistance to surface fouling, and low friction coefficient.7–10 These properties historically made fluorinated surfaces invaluable for multiphase microfluidic chips.11–16 By patterning fluorination amidst a non-fluorinated surface, patterned hydrophobicity has been used to generate droplets, create microarrays, and control microfluidic valving.17–19 Therefore, facile methods to selectively fluorinate the surface of polymeric SLA and DLP resins are required, particularly for the commercially available resins used by most laboratories.

Currently, there are few methods available to generate a fluorinated surface on 3D printed material, particularly a patterned surface. One option is to start directly with a fluorinated resin,20 but these are rare in practice due to limited commercial options. Additionally, fully fluorinated devices are not readily patterned at the surface due to their chemical inertness. Alternatively, selective surface patterning is possible by using printed pieces modified at the surface with fluorinated coatings.6,21,22 Polymeric liquid coatings provide a robust hydrophobic layer up to hundreds of micrometers thick,21 but may be inappropriate for microscale features that are easily blocked or filled in. A chemical vapor deposition method can be used to generate a thin, highly hydrophobic coating by polymerizing a fluorinated acrylate film on the surface, but has limited use in enclosed channels.23,24 Thin coatings can also be achieved by including a polymerization initiator in the resin, to provide covalent anchor points for fluorinated polymer brushes.22 However, polymer brushes may exhibit poor mechanical stability during abrasion.6

Silanization using fluorinated silanes is a reliable method for molecular-scale surface modification of glass and polydimethylsiloxane (PDMS),25,26 but silanization of polymeric materials can be challenging. Historically, polymers have been chemically modified primarily by strategies such as wet etching, plasma or corona treatment, or coatings, rather than direct silanization.27–31 Extensive surface oxidation is usually required to generate enough silane-reactive functional groups (e.g. hydroxyls) at the polymer surface, but not all polymers can withstand such treatment, as they may degrade after plasma exposure.7,27,28,32,33 So far there have been sporadic reports of silanization of resin 3D printed microfluidic devices, e.g. to fluorinate 3D printed molds for PDMS34 and to attach reactive functionalities for bonding of 3D printed pieces.35 In some cases, the printed polymer had to be coated with a layer of silica to enable silanization.36,37 To date, there has been little testing of the conditions required for direct fluoroalkyl silanization of resin printed pieces, nor characterization of the hydrophobicity and stability of the silanized surface.

Here, we aimed to develop a robust and straightforward silanization protocol using (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane, a fluoroalkyl silane, and a suite of commercially available SLA and DLP resins to generate a highly fluorinated surface for use in microfluidic devices. While optimizing the reaction conditions to generate the highest possible contact angle, we found, surprisingly, that surface oxidation using air plasma was unnecessary for silanization. To characterize the surface and investigate reactive groups involved in forming a covalent bond between the printed resin and the fluoroalkyl silane, we measured the air/water contact angle of the silanized surface and used infrared (IR) spectroscopy. We tested the ability of the method to selectively pattern hydrophobic regions in a 3D printed open microchannel, and further tested the applicability of the optimized method to four additional resins. The method is facile, versatile, and allows for dynamic patterning of a hydrophobic surface on a resin-printed piece.

Materials

A Bottle of the Master Mold for PDMS Device Resin

Master Mold Resin

Learn More

Clear Microfluidics Resin V7.0a

Learn More

M50

Experimental Section

3D Printing
Printed parts were designed using Autodesk Inventor 2018. The CAD files were sliced at 50 μM intervals using MII Utility Shortcut V 3.27 and printed using a CADworks3D M50–405 printer (MiiCraft, CADworks3D). The commercial resins included were BV-007A (Clear) (MiiCraft, CADworks 3D), Green Master Mold (MiiCraft, CADworks 3D), Dental LT Clear Resin (V2) (FormLabs), and Asiga PlasClear V2 (iMakr). A house-made photoresin consisting of 0.4 % w/v phenylbis(2,4,6-trimethylbenzoyl)phosphineoxide (Irgacure 819) (Therofisher) dissolved in poly(ethylene glycol) diacrylate (PEG-DA) (MW 250) (Sigma Aldrich) was also included in the suite of resins tested.38 The printer setting for each resin can be found in Table S1. Printed parts were rinsed with either 95% ethanol (Koptec), isopropanol (Fisher chemical), or methanol (Fisher chemical) as recommended for by the manufacturer for the resin. Printed pieces were post-cured in an UV-light box, then stored at room temperature on the bench top in polystyrene petri dishes (Fisher) prior to silanization.

Surface Treatment of 3D Printed Pieces
Where noted, some printed parts were plasma treated using a BD-20AC laboratory corona treater (Electro-Technic Products, Chicago IL, USA). Printed parts were placed 3 mm below the plasma source and treated for 5 – 60 s immediately prior to surface silanization. For gas-phase deposition, 200 μL of neat tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane (Gelest Inc., Morrisville PA, USA) was placed in a vacuum desiccator in a small polypropylene dish, followed immediately by the printed parts, and a vacuum was applied for 2 hours at room temperature. For solvent deposition, the surface of the printed part was submerged in a 10% v/v solution of tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane in solvent (Fluorinert FC-40 (Sigma Aldrich) or 200 proof ethanol (Koptec) for 30 min at room temperature, unless otherwise specified. After silanization, surfaces were rinsed with 95% ethanol and DI water and dried with a nitrogen gun.

Contact Angle Measurement
Surface air/water contact angles were measured using a ramé-hart goniometer (model 200–00, ramé-hart instrument co., Succasunna NJ, USA) and DROPimage Advanced software. Contact angle was measured for 3 separate printed pieces per condition, by pipetting one 5-μL droplet of DI water per print onto the silanized surface. 8×8×8 mm3 cubes were used for the printed piece, and oriented so the smooth flat face of the printed cube was tested.

Surface Chemistry Characterization with Infrared Spectroscopy
The surface chemistry of the printed parts was examined by using an iD7 ATR Nicolet IS5 FT-IR spectrophotometer (Thermo Fischer Scientific). The IR spectrum was measured on the flat smooth face of a 10×10×2 mm3 printed rectangular prism. The instrument was set to a constant gain of 4, and the background was collected prior to each session. Data was collected, visualized, and processed using the OMNIC software (Thermo Fischer Scientific).

Robustness testing
Printed pieces were silanized according to the optimized method. To test the resistance to mechanical damage, the parts were clamped with two binder clips against a clean petri dish to apply constant pressure and rubbed together for 30 s at a time. Air/water contact angles of the silanized surfaces were measured before and after the mechanical test. To test stability after storage, silanized printed parts were stored in a petri dish at room temperature under ambient light, and the air/water contact angles were repeatedly measured over time. Finally, contact angles were measured before and after soaking the printed parts for 2 hours in methanol.

Selective Patterning of 3D Printed Surfaces
Rectangular prisms (20×15×3 mm3) were printed using BV-007A resin. Each print contained an embossed cross-shaped open channel with a rectangular cross-section (1 mm deep, 2 mm wide). Scotch tape (3M) was cut and aligned manually to prevent the fluoroalkyl silane solution from coming into contact with portions of the printed surface inside the channel. Taped pieces were immersed in a solution of 10% v/v (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane in FC-40 for 30 min in a fume hood at room temperature. After treatment, pieces were rinsed with 95% ethanol and DI water and dried with nitrogen. To test the functionality of the patterned surface, solutions of food coloring in water were pipetted into the arms of the embossed features.

3D printed Droplet Generator
A simple T-junction was designed in AutoCAD, consisting of a 10 mm channel with a 0.5 × 0.5 mm cross-section, with a 3 mm channel length with a 0.5 × 0.5 mm cross-section channel that intersects the longer channel. The enclosed channel was fluorinated by filling the channel with a solution of 10% v/v (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane in FC-40 for 30 min, in a fume hood at room temperature. A syringe (1 mL, BD) with a 27 G needle (BD) was filled with FC-40 oil containing 0.5 mg/mL RfOEG (triethyleneglycol mono[1H,1H-perfluorooctyl]ether, a surfactant synthesized in house).29 Another syringe was filled with 1 M Fe(SCN)2+(aq) in water. Connections to the device were made with nonshrinkable PTFE TT-30 tubing (Weico Wire, Edgewood NY, USA). Pressure driven flow was achieved using a Chemyx syringe pump (Fusion 200, Houston TX, USA), using flow rates of 30 μL/min for the oil and 10 μL/min for the aqueous solution. Brightfield images were collected using an Zeiss AxioZoom macroscope (Carl Zeiss Microscopy, Germany) at 1.6 magnification with an Axiocam 506 Mono camera. Images were collected at 1 s intervals for 10 s. All images were analyzed in Zen 2 software.

Data Analysis
Statistical tests and curve fitting were performed using Graphpad Prism version 9. Half-lives and half-times of exponential fits were calculated according to half time = ln 2/k, where k is the rate constant from the fit.

Results and Discussion

Plasma oxidation was not necessary or effective for silanization of SLA printed pieces
While the precise composition of most commercial resins is proprietary, MSDS information states that many are based on acrylate and/or methacrylate polymers (Figure 1a). Silanization of related polymeric materials such as poly(methyl methacrylate) (PMMA) requires oxidation to generate hydroxyl groups that undergo condensation reactions with the silane reagent.7,14 Similarly, prior reports of silanization of an acrylate-based 3D printed material included activation of the surface with plasma treatment.34,35 Therefore, we first tested the efficacy of silanization of 3D printed pieces as a function of the duration of exposure to air plasma. As a case study, we selected a clear (meth)acrylate-based resin formulated specifically for printing microfluidic devices, BV-007A resin from MiiCraft, and sought to silanize it with (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane (Figure 1a). Two common methods of silanization were tested: gas-phase deposition29,34,39–41 and liquid-phase deposition.14,29 For the latter, we used a 10% v/v solution of silane in FC-40 fluorinated oil.t.

Figure 1: Effects of plasma treatment and silanization on the chemistry and hydrophobicity of DLP printed pieces. (a) Chemical structures of the fluoroalkyl silane and monomer acylate and methacrylate base used for many resin formulations. (b) Air/water contact angles of BV-007A after silanization by solution-phase (blue squares, FC-40 solvent) or gas-phase (pink dots) deposition after varied times of treatment with air plasma (n=3, mean ± std dev). The black triangle represents printed BV-007A pieces that received neither plasma treatment nor any silane treatment. Two-way ANOVA for solution vs gas-phase silanization (**** p<0.0001). (c) ATR-FT IR spectrum of the BV-007A surface with no exposure to air plasma (pink) and after 30 s plasma treatment (grey), without silanization. Spectra are offset to display spectral features. (d) Air/water contact angles of BV-007A surface after solution-phase silanization in FC-40 (pink dot) or ethanol (blue square). Two-way ANOVA with Sidak’s multiple comparisons to compare between solvents (****p <0.0001, *** p<0.001).

Surprisingly, we found that even in the absence of plasma treatment (0 s exposure), silanization significantly increased the air/water contact angle for both methods (gas phase, p<0.005; solvent, p<0.001) compared to the 60° contact angle of the unslianized printed piece angle (Figure 1b, Figure S1). While gas-phase deposition provided a contact angle near 90°, the lower boundary for hydrophobicity, the solution-phase method provided a significantly larger (p < 0.0001) contact angle close to 120°, the upper limit for a flat, fluorinated surface.26,42,43 Plasma treatment from 5 to 60 s did not further increase the contact angle. Wanting to further test the impact of plasma cleaning on the surface chemistry, we next used IR spectroscopy to investigate functional groups on the surface of printed BV-007A pieces.

We expected that sufficient exposure of BV-007A pieces to air plasma would oxidize the surface to form alcohol and/or carboxylic acid groups.44 To characterize the surface chemistry and investigate the extent of surface activation at short plasma treatment times, we collected surface ATR FI-IR spectra of the printed pieces (Figure 1c). As expected for (meth)acrylate-based BV-007A, the spectra closely resembled that of a commercial sheet of PMMA (Figure S2). The peaks at 2970, 2930, and 2870 cm−1 were assigned to alkane sp3 C-H stretching. A major C=O stretch peak at 1718 cm−1 was attributed to the carbonyl in the backbone of the (meth)acrylate-based polymer as well as other carbonyl-containing components of the resin, e.g. photoinitiators and photoabsorbers. The C-O-C stretching was assigned to the peaks ranging from 1000 – 1300 cm−1 in the fingerprint region.45 Treating BV-007A printed pieces with air plasma for 30 – 60 s did not alter the IR spectra substantially (Figure 1c and data not shown). In particular, no characteristically broad alcohol band (3550 – 3200 cm−1) was observed, and there was no change in the alkyl CH stretches or carbonyl peak. These data were consistent with plasma treatment neither affecting the contact angle of the material (Table S2) nor improving its silanization (Figure 1b). As a positive control, oxidation from the plasma treatment was verified using both glass and PDMS, whose contact angle decreased after 5 s of plasma treatment as expected (Table S2). Prior reports of plasma treatment of PMMA used longer treatment times (5 min and greater) to modulate the surface polarity,46,47 but we found that treatment of BV-007A pieces with air plasma for longer than 2 min generated cracks in the surface. Since plasma treatment was unnecessary for silanization and in fact was ineffective at oxidizing the BV-007A surface at short times, we proceeded to optimize and characterize the silanization of BV-007A pieces it its absence.

Solvent deposition was most effective when a fluorocarbon oil was used as a solvent.

Having established that solution-phase deposition was more effective than gas-phase deposition, we further optimized the choice of solvent and concentration of silane. Two solvents were tested: ethanol (200 proof), a common solvent for deposition of trichlorosilanes,29,30 and FC-40, a fluorinated oil.14 Whereas deposition from ethanol solution was largely ineffective (contact angles < 90°) regardless of silane concentration, deposition from FC-40 solution had a concentration-dependent effect, yielding an average contact angle of ~ 120° at 10 % v/v silane (Figure 1d). Exposure to FC-40 alone, without silane, did not increase the contact angle significantly (Figure S1). Therefore, 10% v/v of the fluorinated silane in FC-40 was used for all further experiments.

Time dependence of the reaction provides support for covalent bond formation

Next, we tested the time dependence of the silanization reaction. The contact angle increased in a time-dependent manner with a half-time of 3.4 min, reaching a plateau after 15 min that was unchanged for the rest of the testing period, up to 60 min (Figure 2a). To complement the contact angle data and assess the extent of bond formation between the fluoroalkyl silane and BV-007A, ATR-FT IR spectra were collected from these samples (Figure 2b). The spectra changed noticeably over this time period. In particular, the carbonyl stretch at 1716 cm−1 decreased in intensity over time (Figure 2b – c), and the peak area was well fit by exponential decay equation with a half-life of 3.5 min (Figure 2d). This observation suggested a molecular reaction between the resin and the fluoroalkyl silane that consumes a carbonyl. The data do not distinguish between the methacrylate carbonyl and any carbonyls that may be present in the resin’s photoinitiators or photoabsorbers, but these additives typically are a minor constituent of the resin. An immediate increase in fingerprint region intensity was consistent with the addition of fluoroalkyl silane to the surface of the print (Figure 2b, ,cc and ande).e). New peaks included those at 1023 cm−1, assigned to Si-O-R stretching,48 1232 and 1142 cm−1, consistent with asymmetric and symmetric C-F stretches, and 707 cm−1, assigned to the C-F wag.49 This increase had a half-time of only 2.0 min, shorter than the decay of the carbonyl, suggesting that physical adsorption of the silane may have preceded the covalent reaction (Figure S3). The -C-H stretch peaks at 2872, 2932, and 2971 cm−1 were still present after silanization (Figure 2b – c).50

Figure 2: Time dependence of the chemical reaction. (a) Contact angle of the fluorinated surface after various amount of silane treatment (n=3, mean ± std dev). The data were fit to an exponential curve, y = 111 – 46.9e−0.201x, R2 =0.844. Insets show images of droplets on BV-007A surface after 0 and 30 min of silanization. (b) ATR-FT IR spectrum of printed BV-007A pieces after various times of silane treatment. Two regions of interest are highlighted: the carbonyl peak at 1720 cm−1 and the finger print regions 650–1300 cm−1. (c) The chemical structures present in a methyl methacrylate-based resin and from the fluoroalkyl are labeled with the corresponding IR spectra peak. (d) The area under the carbonyl peak decreased in a time-dependent manner (n=3, mean ± std dev), fit to an exponential decay y = 55.9e−0.115x +43.3, R2 = 0.936. (e) The area under the curve of the finger print region increased in a time-dependent manner, fit to an exponential curve, y = 57.7 – 24.2e−0.348x, R2 =0.855.

From both the contact angle measurements and the IR spectra, we concluded that the silanization reaction likely resulted in a covalent bond, and that 30 min was sufficient for reaction completion and generation of a highly hydrophobic surface. We note that the mechanism for such a reaction does not match that of typical silanizations, which occur through a condensation reaction with hydroxyl groups on the surface of the material. In this case, there were no detectable hydroxyl groups, yet surface modification still occurred. We were unable to find a precedent for the reaction of a tri-substituted silane with (meth)acrylate; the closest reaction we found in literature was that of silyl radicals attacking alkenes and acrylates,51 but we would not expect formation of a silyl radical in this system.

Robustness and stability of fluorination procedure

To establish the practical utility of the method, we considered the sensitivity of the procedure to the state of the printed piece and characterized the stability of the hydrophobic surface. First, we considered that the surface chemistry of the printed piece may change over time and potentially alter the reactivity with the trichlorosilane, e.g. due to slow cross-linking of residual monomer under ambient light.52,53 To test the efficacy of silanization as a function of light-induced aging, printed parts were treated with either the manufacturer-recommended 20 s or an extended 360-s UV exposure during the post-curing process. We estimate that continuous 360-s exposure was an equivalent dose of light as being on a bench top under ambient light for 32 days (Table S3). The extended UV cure created discoloration and warped some of the pieces, so only pieces with a flat top surface were used for subsequent silanization. No significant difference was observed in the water contact angles of the control pieces (20 s) compared to the pieces with extended UV exposure (360 s), either before or after silanization (Figure 3a). This result was consistent with our informal observations that month-old BV-007A pieces yielded similar contact angles after silanization as recently printed (1–3 days old) pieces. Therefore, the silanization method appears insensitive to the age of the piece, at least in this timescale, which enables robust fabrication procedures.

Figure 3: Robustness of the method to the age of the printed piece, abrasion, and storage time after silanization. (a) Contact angle of DLP printed pieces (BV-007A) that were silanized with or without extended UV curing (n=3 printed parts for each condition, mean ± std dev). Two-way ANOVA with Tukey’s multiple comparisons (ns, p>0.05, ** p<0.005). (b,c) Contact angle of silanized BV007 after (b) deliberate mechanical abrasion under constant pressure or (c) long term storage. n=3 printed parts for each condition, mean ± std dev. Some error bars too small to see. One-way ANOVA (ns, p>0.05).

Next, we assessed the robustness of the silanized surface when subjected to mechanical damage and extended storage, a property that affects the range of potential uses, handling, and storage. Microfluidic chips must be able to withstand mild abrasion during the movement of the device, and in particular we anticipated using this method to generate fluorinated SlipChips, which rely on sliding parts past one another.11 Therefore, silanized printed pieces were subjected to gentle mechanical damage by manually rubbing the piece against a clean polystyrene surface, mimicking normal wear and tear during use. The water contact angle of the fluorinated pieces of BV-007A was not significant altered by this process (Figure 3b), indicating that the surface is stable under mild abrasion conditions. Similarly, when silanized pieces of BV-007A were stored on the bench, the contact angles remained unchanged for at least 154 days, the longest time point measured (Figure 3c). We did observe that the initial contact angle in these experiments was slightly lower than in previous experiments, which we attribute to hydrolysis of the trichlorosilane during storage because replacement of the silane stock improved the hydrophobicity (data not shown). We concluded that the silanized surface was quite stable and the method was robust to the age of the resin though sensitive to the quality of the silane stock, all of which are consistent with the formation of a covalent bond during the silanization reaction.

Patterning of surface hydrophobicity on 3D printed parts

Compared to printing with a fully fluorinated resin, site-specific patterning is an advantage of post-print modifications, offering the potential for passive fluidic control. Therefore, we tested the ability of the silanization protocol to selectively pattern hydrophobic patches on the surface of BV-007A resin, using a pair of intersecting open channels in a simple, recessed cross design. The arms of the cross were protected from the silanization using adhesive tape, while the center square was silanized to generate a pattern of four separate fluid compartments, separated by a surface tension barrier. In the non-silanized control, colored solutions pipetted into the arms of channel mixed readily in the center of the cross (Figure 4a, Not Treated), whereas a micropatterned hydrophobic patch in the center of the cross successfully constrained the solutions to the arms (Figure 4a, Pattern). These data demonstrate that because the silanization method requires contact of the liquid silanization solution with the printed surface, it is easily patterned by physical masking strategies to define the silanized area.

Figure 4: Application of the optimized silanization procedure for surface patterning and droplet generation. (a) Selective surface patterning of an open channel. Parts printed in BV-007A were patterned so that the center of the cross was hydrophobic. In a non-silanized piece (top), the blue and yellow food dyes mixed in the center; in the piece patterned by local silanization (bottom), the droplets remained distinct from each other. The width of the channels in these photos was 2 mm. (b) Fluorination of 3D printed fluidic channels in a T-junction droplet generator. (Top row) Photos of empty 3D printed chip at low and high magnification, with inlets and outlet marked. (Bottom row) Images of two-phase fluid flow with and without silanization of the interior channels. The dark liquid is the aqueous solution; the fluorinated oil is colorless. Droplets formed only in the silanized system. Scale bar 1 mm.

Silanization of an enclosed channel for droplet formation

In addition to surface features, many microfluidic devices feature enclosed channels whose surface chemistry must be controlled, e.g. to present a fluorophilic interior surface for droplet microfluidics.54 We reasoned that the liquid-based silanization described here to could be used to fluorinate enclosed channels by simply filling the channel with silanization solution for 30 min and rinsing afterwards; no prior surface oxidization is needed. To test this prediction, a simple T-junction droplet generator was printed (Figure 4b) and used to create water-in-oil droplets using an aqueous solution of Fe(SCN)32+ and fluorinated oil. As expected, when the 3D printed device was not silanized, the aqueous solution wetted the channel and droplets did not form (Figure 4b, Not Treated), whereas fluorosilanization by this protocol prevented wetting and enabled the formation of droplets (Figure 4b, Silanized).

Silanization of a suite of SLA resins demonstrates broad applicability

This silanization protocol would be most useful if applicable across a variety of SLA and DLP resins. Therefore, in addition to BV-007A, we tested three commercially available resins: Dental (FormsLab), Green Master Mold resin (CADworks3D), and Plasclear (iMakr), plus a polyethylene glycol diacrylate (PEG-DA)-based resin developed by the Folch laboratory.38 The resins chosen come from 3 different companies and span a breadth of applications (PDMS mold, dentistry, small part/figurine prints) and properties (low viscosity/high resolution, biocompatible, heat stable) of interest to microfluidic device fabricators. Based on our prior data that the extent of reaction correlated with diminished absorbance from the carbonyl in the IR spectrum, we hypothesized that any resin with an acrylate (BV007A and PEG-DA) or methacrylate (Dental, Green Master Mold, and Plasclear) backbone, or possibly with carbonyl-containing photoinitiators or photoabsorbers, would react with the fluoroalkyl silane. Following the optimized protocol, all printed pieces were submerged in a 10% (v/v) solution of fluorinated silane in FC-40 oil for 30 min, without plasma treatment. This procedure successfully increased the contact angle for each material compared to its non-treated control (Figure 5a). The Green Master Mold, Dental, and BV007 resins were highly hydrophobic after silanization, with contact angles of ~ 115–118°. In contrast, the PEG-DA and Plasclear resins had a mildly hydrophobic contact angle, near 100°. This trend was reproduced in two independent experiments.

Figure 5: Testing the optimized silanization method with a variety of SLA/DLP resins. (a) Air/water contact angles for resins prior to silanization, after silanization with the optimized procedure, and after soaking in methanol (n=3 printed parts for each condition, mean ± std dev). Two-way ANOVA with Tukey’s multiple comparisons tests (ns, p>0.05, ** p<0.005). (b) ATR FT-IR spectra of pieces printed using the various resins, before and after silanization (n=3 printed parts for each condition). Spectra are offset to display spectral features.

To test the extent to which the high contact angles may have been due to physically adsorbed silane, the silanized pieces were soaked for 2 hours in methanol.55 First, the efficacy of methanol to remove a significant fraction of physically adsorbed silane was confirmed on a piece of acrylic (Figure S4). Next, silanized 3D printed parts were tested; in all cases, the hydrophobic surface persisted, again suggesting that the silane was covalently bound (Figure 5a). Surprisingly, the contact angle increased for both the Plasclear and PEGDA resins, an observation that remains to be explored. While here we tested stability to methanol treatment, we recommend that the stability of the fluorinated surface be tested under the experimental conditions relevant to the intended use of the printed piece, e.g. under varied pH or solvents.

We examined the surface chemistry of the printed pieces by ATR FT-IR to potentially explain the difference in susceptibility to silanization between resins (Figure 5b). The pair of peaks at 1453 and 1510 cm−1 are useful to distinguish PMMA from poly(methyl acrylate) (PMA).56 The 1453 cm−1 peak, which was present in all samples, is attributable to a methylene vibration -CH2- found in both PMMA and PMA. The peak at 1510 cm−1, attributable to the methyl vibration C-CH3, is indicative of PMMA. This peak was present in all four commercial resins tested, suggesting the presence of methacrylates in these materials; as expected, it was not seen in the PEGDA sample. A small peak at 1630 cm−1, assigned to C=C bonds from residual (meth)acrylate monomers,57 was present in all samples.

Next, changes in the surface IR spectra after silanization were examined. The three resins that exhibited a larger change in contact angle after silanization (BV-007A, Green Master Mold, and Dental) also showed a larger decrease in the intensity of the carbonyl peak at 1719 cm−1 (Figure 4c). Furthermore, the decrease in the carbonyl peak correlated with appearance of peaks consistent with deposition of the fluoroalkyl silane. The peaks at 1232, 1142 and 707 cm−1 were again assigned to stretches and wagging of the fluoroalkyl chain,23,49 and they increased after silanization for the four commercial resins. Similarly, the peak at 1010 cm−1 that increased after silanization in the commercial resins may be a part of the Si-O-R stretch (usually a strong and broad stretch, 1000–1100 cm−1).58 In contrast, Plasclear and PEGDA, which had smaller changes in contact angle, showed less consumption of the carbonyl, and PEGDA showed no increase in the finger print region. From these data, we concluded that while all five resins showed an increase in contact angle that was resistant to removal by methanol, only a fraction of them formed a covalent bond that consumed a carbonyl. It may be significant that PEGDA, which has no added photoabsorbers, was the least reactive of the materials towards the silane; this possibility was not tested further here. Go to:

Conclusions

We have demonstrated a robust and versatile strategy to control the surface chemistry and hydrophobicity of DLP 3D printed parts by reacting the printed surface with an alkyl-fluorinated silane. The optimized method consisted of simply placing a DLP-printed part directly into a solution of 10% v/v (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane in FC-40 for 30 min, then thoroughly rinsing the part with ethanol and water and drying under nitrogen. This method did not require any pre-treatment of the printed piece. The reaction between the silane and the resin appeared to consume a carbonyl present in the resin material, and was consistent with covalent bond formation by an unknown mechanism. This method created a hydrophobic surface with air/water contact angles close to 120 deg. Additional work would be needed if superhydrophobicity (> 150°) was required, e.g. by adding multi-scale surface roughness to the printed surface prior to silanization.59 The fluorinated surface was resistant to mechanical damage, methanol soaking, and 154 days of storage, and the method was compatible with printed parts even after significant light exposure. Selective patterning of a hydrophobic surface was demonstrated in 3D printed open channels by a simple masking method. Furthermore, the method was effective with a suite of (meth)acylate based resins, with higher contact angles correlating with greater consumption of the carbonyl. We anticipate that simple approach to controlling the surface chemistry of resin 3D printed microfluidic parts, including for selective fluorination of specific regions, will advance the fabrications of complex two-phase devices and enable greater control of the wettability of 3D printed parts.

Design and characterization of a 3D-printed staggered herringbone mixer

by

Design and characterization of a 3D-printed staggered herringbone mixer

Vedika J Shenoy , Chelsea ER Edwards , Matthew E Helgeson  & Megan T Valentine

3D printing holds potential as a faster, cheaper alternative compared with traditional photolithography for the fabrication of microfluidic devices by replica molding. However, the influence of printing resolution and quality on device design and performance has yet to receive detailed study. Here, we investigate the use of 3D-printed molds to create staggered herringbone mixers (SHMs) with feature sizes ranging from ∼100 to 500 μm. We provide guidelines for printer calibration to ensure accurate printing at these length scales and quantify the impacts of print variability on SHM performance. We show that SHMs produced by 3D printing generate well-mixed output streams across devices with variable heights and defects, demonstrating that 3D printing is suitable and advantageous for low-cost, high-throughput SHM manufacturing.

We kindly thank the researchers at University of California for this collaboration, and for sharing the results obtained with their system.

Method Summary

We investigate the use of 3D printing to create staggered herringbone mixers (SHMs) and show that such devices generate well-mixed output streams across devices with variable heights and defects. This demonstrates that 3D printing is suitable and advantageous for low-cost, low-effort, high-throughput micromixer manufacturing.

Keywords: 3D printer,calibration 3D printing, microfluidics,micromixers, staggered herringbone mixer

Microfluidic mixers such as the staggered herringbone mixer (SHM) [1] promote eddy-like mixing of laminar flow streams, avoiding prohibitively long channel lengths and enabling applications in drug delivery and discovery [2], chemical synthesis [3], sample concentration [4] and biological analysis [5–7]. In SHMs, asymmetric herringbone grooves embedded in the floor or ceiling of the rectangular channel cause transverse flow within fluid streams, promoting mixing by increasing local vorticity (Figure 1) [1,8]. Alternating the grooves' offset between cycles increases flow irregularity, further mixing the two streams [9]. The SHM's mixing proficiency and mechanism has been characterized extensively [8,10] due to its efficiency and ease of design compared with other grooved micromixers [11]. Their low shear flow properties [6] and ability to circulate flow within the channel [4] make SHMs particularly advantageous for biomedical applications,

h/H ∼ 0.36, w/d ∼ 0.45, and a ∼ 95°. See Supplementary Materials for additional information.

Despite these advantages, the widespread use of SHM devices for on-chip diagnostics [4,6,12] is hindered by the reliance on time-intensive photolithography-based fabrication to generate molds for polydimethylsiloxane (PDMS)-based microfluidic devices. Photolithography requires cleanroom training and costly equipment and reagents [13]. Consequently, 3D printing is emerging as an attractive substitute for photolithography due to its comparative affordability, simplicity [14], and the ease of fabricating multilevel designs [13]. 3D printing also enables rapid prototyping by reducing typical fabrication times from days to hours [14].

To establish the use of 3D printing to generate molds for PDMS-based microfluidic devices such as the SHM micromixer, experimental validation of printer and device performance is required. For current 3D printers [15], the minimum feature size is significantly larger than that provided by photolithography, print-to-print variability is expected and random disfigurations can occur, all potentially influencing device performance. Moreover, there are often discrepancies between the targeted and actual dimensions of the 3D printed parts [16]. Unfortunately, not only is information concerning such limitations generally unavailable for any given 3D printer model, but the lack of reporting on chip-to-chip variability is a known barrier for large-scale manufacturing of microfluidic devices [17]. Thus, understanding how print quality influences pattern transfer and mixing performance is especially important for low-cost, low-effort manufacturing that maintains efficacy, as required for microfluidic diagnostic platforms.

To experimentally investigate print quality, we printed a series of raised channels, with heights and widths ranging from ∼25 to 700 μm, using a Miicraft Ultra 50 3D digital light processing (DLP) printer, with a horizontal and vertical resolution of ∼30 μm and ∼5 μm, respectively, and using Resin Works 3D Master Mold Resin for PDMS. We sized the 3D-printed parts by imaging with a Keyence VHX-5000 microscope (Supplementary Figure 1), and found them to be consistently smaller than the programmed design dimensions, with greater discrepancies in height than in the planar dimensions. These differences were consistent across all printed molds, so the programmed design dimensions could be calibrated to achieve printed parts with the correct size (Supplementary Figure 2).

This calibration critically informed mold design, which was performed in Solid Works. Features with design dimensions ≥120 μm (∼100 μm actual dimension) were reproducibly printed, whereas features with smaller design dimensions were often deformed or occasionally missing. We then printed a series of herringbone grooves with a fixed width w of 100 μm and herringbone spacing d that varied from 50 to 500 μm (Figure 1). Printed grooves with designed d < 300 μm (∼220 μm real spacing) frequently fused with the adjacent herringbones, with increasing severity for decreasing d (Supplementary Figure 1A & B). We note that these disfigurations existed within the molds themselves, not only the PDMS replicates. Thus, we established a minimum feature size of 100 μm and fixed d = 300 μm in the design of SHM molds for further study.

These constraints required SHM features in 3D-printed molds to differ somewhat from the geometries recommended by prior work [18], which find mixing to be most effective near w/d ∼1 and w = 50 μm. By contrast, our mixer has w/d ∼ 0.45 and w = 100 μm. However, we achieved a ratio of herringbone height h to channel height H, h/H ∼ 0.36, and herringbone angle a ∼ 95°, the latter of which was informed by computational work that suggested that angles in the range of ∼90°–105° produce optimal transverse flow [1,19]. Ten herringbones per each half cycle were included in accordance with prior experimental studies [10]. To compensate for the larger feature sizes and spacing requirements of 3D-printed molds and prevent the microchannel length from being excessively long, we limited the number of mixing cycle elements to five on a single chip (total channel length ∼41 mm) (Supplementary Figure 3). We then investigated a range of flow rates to produce well-mixed streams with these larger dimensions.

To demonstrate the mixing capabilities of PDMS-based micromixer devices generated using the 3D-printed SHM molds, we examined the mixing performance of two exemplar devices generated with two identically designed, but individually printed, molds. In detail, PDMS, obtained as commercially available Sylgard 184 (Dow Chemical, MI, USA), was mixed at a 10:1 ratio of PDMS to cross-linker, poured in the treated molds, degassed for approximately 2 h, then cured at 60–80°C for 4 h and removed from the mold. We did not observe qualitative deviations in the PDMS replicates formed from either mold, consistent with the well-established high-fidelity pattern transfer from 3D-printed molds to PDMS [20]. Each device was tested at flow rates from 0.5 to 20 μL min-1, corresponding to Peclet numbers Pe = 200–8000, a useful range for low shear-stress mixing of biological materials [21]. The Peclet number describes the ratio of advective to diffusive transport and is given by Pe = uL/D where u is the average velocity defined by u = q/WH and q is the volumetric flow rate, L is a characteristic length scale given here by W/2 and D is the Stokes–Einstein diffusivity.

For each test, a 0.30 μM aqueous solution of 70 kDa fluorescein isothiocyanate (FITC)-dextran was delivered into Input 1, while neat water was delivered into Input 2 at an equivalent flow rate. The mixing of the two streams under steady-state flow conditions was observed via fluorescence microscopy (see Supplementary Materials for details). Representative intensity maps of the flow fields along a cross-section of the channel after one, three and five mixing cycles are shown in Figure 2. For quantitative analysis, the intensity was normalized by dividing by the maximum value (Supplementary Figure 4), and its coefficient of variation (CV), defined as the standard deviation divided by the mean, was calculated (Figure 3 & Supplementary Figure 5). We consider any profile with a CV < 0.1 to be well mixed [10].

The color scale indicates the presence of fluorescein isothiocyanate (FITC)-dextran, introduced through Inlet 1 and mixed with neat water introduced through Inlet 2. The dark blue regions indicate low fluorescence signal (i.e., mostly water), whereas red regions indicate high FITC signal; uniform light blue/green regions indicate homogeneous mixing. At Pe = 200 streams appeared homogeneous and well-mixed after three cycles, while at Pe = 8000 striations remained even after five cycles.

Devices generated using 3D-printed SHM molds demonstrate good mixing performance. The number of cycles required to achieve good mixing depends on Pe, as expected. At Pe = 200, where diffusion is prominent, only three mixing cycles are required to achieve CV < 0.1, and only four to five mixing cycles are needed at Pe values of 800–6000 (Figure 3). We could not achieve good mixing for Pe > 6000 with this design due to the limited channel length. We observed a modest decline in mixer performance compared with that of devices with smaller feature sizes achieved via photolithography, where it is possible to achieve CV < 0.1 with two mixing cycles at Pe = 625 and with five cycles at Pe = 6250 [10].

In both cases well mixed streams with CV < 0.1 black dashed line) are achieved for Pe < 6000. Error bars were calculated via propagation analysis as described in the Mixing Analysis section in the Supplemental Methods.

Finally, we examined the effects of print-to-print variation on mixing by comparing the results obtained using two devices containing chipped herringbones, with the second device containing twice as many defects as the first, and modest differences in herringbone properties (Supplementary Tables 1–4). Device 2 exhibits larger CV values across almost all Pe values after one and two cycles of mixing; however, these differences vanish after three to five cycles of mixing (Supplementary Figure 5). That print-to-print variability does not compromise performance suggests that rigorous measurements of each mold are not necessary beyond the initial characterization of 3D printer capabilities and 3D printing can deliver SHM devices that yield reproducible results over multiple prints, even in the presence of defects.

In summary, we introduced methods to characterize and calibrate 3D printer outputs and adapt the SHM geometry to enable 3D printing of replica molds. We demonstrated good mixing performance despite the modified dimensions and in the presence of print-to-print variation and defects. This method provides significant advantage for applications that benefit from rapid, low-effort manufacturing.

Apparatus

Master Mold for PDMS Resin

Master Mold for Resin

Learn More

M50

Learn More

3D Printing of Inertial Microfluidic Devices

by

3D Printing of Inertial Microfuidic Devices

Sajad Razavi Bazaz, Omid Rouhi, MohammadAmin Raouf, Fatemeh Ejeian, MohsenAsadnia, Dayong Jin and Majid Ebrahimi Warkiani

Inertial microfluidics has been broadly investigated, resulting in the development of various applications, mainly for particle or cell separation. Lateral migrations of these particles within a microchannel strictly depend on the channel design and its cross-section. Nonetheless, the fabrication of these microchannels is a continuous challenging issue for the microfluidic community, where the most studied channel cross-sections are limited to only rectangular and more recently trapezoidal microchannels. As a result, a huge amount of potential remains intact for other geometries with cross-sections difficult to fabricate with standard microfabrication techniques. In this study, by leveraging on benefits of additive manufacturing, we have proposed a new method for the fabrication of inertial microfluidic devices. In our proposed workflow, parts are first printed via a high-resolution DLP/SLA 3D printer and then bonded to a transparent PMMA sheet using a double-coated pressure-sensitive adhesive tape. Using this method, we have fabricated and tested a plethora of existing inertial microfluidic devices, whether in a single or multiplexed manner, such as straight, spiral, serpentine, curvilinear, and contraction-expansion arrays. Our characterizations using both particles and cells revealed that the produced chips could withstand a pressure up to 150 psi with minimum interference of the tape to the total functionality of the device and viability of cells. As a showcase of the versatility of our method, we have proposed a new spiral microchannel with right-angled triangular cross-section which is technically impossible to fabricate using the standard lithography. We are of the opinion that the method proposed in this study will open the door for more complex geometries with the bespoke passive internal flow. Furthermore, the proposed fabrication workflow can be adopted at the production level, enabling large-scale manufacturing of inertial microfluidic devices.

We kindly thank the researchers at University of Technology Sydney for this collaboration, and for sharing the results obtained with their system.

Introduction

Inertial microfuidics has been broadly investigated, resulting in the development of various applications, mainly for particle or cell separation. Lateral migrations of these particles within a microchannel strictly depend on the channel design and its cross-section. Nonetheless, the fabrication of these microchannels is a continuous challenging issue for the microfuidic community, where the most studied channel cross-sections are limited to only rectangular and more recently trapezoidal microchannels. As a result, a huge amount of potential remains intact for other geometries with crosssections difcult to fabricate with standard microfabrication techniques. In this study, by leveraging on benefts of additive manufacturing, we have proposed a new method for the fabrication of inertial microfuidic devices. In our proposed workfow, parts are frst printed via a high-resolution DLP/SLA 3D printer and then bonded to a transparent PMMA sheet using a double-coated pressure-sensitive adhesive tape. Using this method, we have fabricated and tested a plethora of existing inertial microfuidic devices, whether in a single or multiplexed manner, such as straight, spiral, serpentine, curvilinear, and contraction-expansion arrays. Our characterizations using both particles and cells revealed that the produced chips could withstand a pressure up to 150psi with minimum interference of the tape to the total functionality of the device and viability of cells. As a showcase of the versatility of our method, we have proposed a new spiral microchannel with right-angled triangular cross-section which is technically impossible to fabricate using the standard lithography. We are of the opinion that the method proposed in this study will open the door for more complex geometries with the bespoke passive internal fow. Furthermore, the proposed fabrication workfow can be adopted at the production level, enabling large-scale manufacturing of inertial microfuidic devices.

Continuous separation of particles and cells is required for a wide variety of applications that include mineral processing, chemical syntheses, environmental assessments, and biological assays1 . A number of conventional methods exist for this purpose; however, they have several drawbacks. Membrane fltration-based techniques, while effici t and simple, are limited by flter fouling and clogging. Centrifugation methods are also plagued by problems of particle adhesion and clogging, along with their high cost and inability for continuous processing. Likewise, techniques based on sedimentation are prone to particle adhesion and slower processing time, which increases the non-viability of cells in biological applications. Also, methods based on magnetic-activated cell sorting (MACS) and fuorescence-activated cell sorting (FACS) are proven to be low throughput and expensive2–4 . With the evolution of microfabrication and rapid prototyping techniques, microfluidic technology has emerged as an alternative to improve upon conventional separation techniques5,6 . Tese microfuidic techniques are grounded on the unique characteristics of microscale fl w phenomena and have recently gained prominence as effici t tools for the control and focusing of microbeads. Amongst existing microfuidic systems, inertial microfuidics has experienced massive growth in many applications ranging from cell separation7,8 , cytometry9,10, multiplexed bio-assays11,12, and also fuid mixing13. Despite great advantages of inertial microfuidics, the commercial impact and scalability of this technology have been restricted due to fabrication issues. As a passive technique, inertial microfuidic systems manipulate cells and particles by taking the advantage of hydrodynamic forces in microchannels with a variety of cross-sections. To date, several microchannels (i.e.,straight, spiral, and serpentine) with diferent cross-sections (i.e., square, rectangular, triangular, trapezoidal, and circular) have been proposed to enhance particle sorting by optimizing the synergetic efects of inertial and Dean drag forces14–18. Tese devices are mainly fabricated by casting PDMS on a master mold, which is made by either standard microfabrication techniques (i.e., silicon etching or SU8 lithography) or using conventional micromilling on an aluminum or polymethylmethacrylate (PMMA) sheets19–21. While this approach has been the workhorse behind the development of majority of these devices, the inability to build non-orthogonal and non-planar structures, cost, and labor intensiveness of the process have hampered its widespread applications and commercialization22,23.

Besides the aforementioned approach, other groups attempted to develop alternative strategies for the fabrication of inertial microfuidic devices. For instance, several groups reported the usage of femtosecond laser irradiation and CO2 laser ablation techniques to produce straight and spiral shape microchannels inside a glass or PMMA24–28. Despite the simple fabrication process, the complexity of building non-rectangular cross-sections, poor surface fin sh, and lengthy etching steps are making them less user-friendly. Some groups also proposed the utilization of metal micro-wires or a sacrific al template in conjunction with soflithography to produce inertial microfuidic devices29–31. In spite of the simplicity of this method in the fabrication of circular channels, PDMS rupture or distortion and the presence of residuals in microchannels during the template removal restrict its utility.

Recently, the fabrication of PMMA microchannels using hot embossing technique has also been reported. While this method is attractive for rapid prototyping and high volume production of microfuidic systems with microscale features, the necessity of using sophisticated equipment limits its widespread usage32,33. Recently, additive manufacturing has emerged as a powerful platform to fabricate 3D functional microfuidic systems from a variety of polymeric materials34. Th s outstanding technology enables investigators to build microstructures with complex shapes and geometries in a short time35,36. Benefting from the stereolithography apparatus (SLA) technique37, Lee et al. directly fabricated a 3D helical trapezoidal microchannel to separate E. coli bacteria using magnetic nanoparticle clusters38. However, due to the poor transparency of the fabricated channel, imaging (whether fuorescent or bright fi ld) was not feasible through the channel. Besides, to remove residuals from channels, the channel width is in the order of millimeter-sized dimensions, which is not suitable for most of the inertial microfuidic applications where small cells or particles are of interest. More lately, 3D printing of sacrific al molds combined with soflithography has gained signifi ant attention due to its simplicity and cost-efectiveness39. Gaining the effici cy of the fused deposition modeling (FDM) printer, Tang and colleagues40 fabricated various microchannels with unconventional cross-sections to study the efect of geometry on elasto-inertial focusing. While this approach is suitable to fabricate microchannels with diferent cross-sections, the resolution of printed parts is not high enough due to inherent limitations of FDM printing. Although direct fabrication of microchannels using SLA and digital light processing (DLP) method is a suitable candidate, inertial microfuidic devices ofen operate in channels in the order of micrometer (e.g., rectangular with 200 µm width and 40µm height) where removing resin residuals from the channel is a challenging issue41.

To address these inadequacies, we have developed a robust protocol for large-scale manufacturing of inertial microfuidic systems. Tanks to the capabilities of DLP and SLA 3D printing42,43, we have printed a wide range of microchannels with diferent geometries, capable of performing particle and cell focusing for various Reynolds numbers (Re). Te approach makes the use of a double-coated pressure-sensitive adhesive tape that perfectly binds open 3D-printed microchannels with optically transparent acrylic sheets, producing a leakage-free interface for inertial microfuidic applications. Te bonding strength is quantifi d, and the compatibility of the concept for the fabrication of new generation of inertial microfuidic devices is evaluated using cells and particles.

Materials

Clear Microfluidics Resin V7.0a

Learn More

M50

Learn More

Results and discussion

Fabrication and characterization of 3D-printed channels.
Soflithography using PDMS and a master mold is a frequently used method for the fabrication of microfuidic systems. Th s strategy has several advantages; for instance, PDMS is biocompatible, optically transparent, and gas permeable, which makes it suitable for a myriad of biological applications44. Also, 3D printing of PDMS has been reported using stereolithography approach45. However, certain drawbacks such as lack of chemical stability, deformation under pressure, and adsorption of small hydrophobic molecules have hindered its industrial-scale utilization. Moreover, the manual molding, cleaning, and bonding process complicate the mass production.

Although theoretically straightforward, scaling up of PDMS-made microchannels for commercialization application in inertial fl ws is challenging since these devices are flex ble and prone to rupture or collapse at high fl w rates. In addition, infation and hysteresis feature of PDMS create a big question mark regarding the exact focusing position of particles. Tis becomes more serious in CFD modeling where the “fxed wall boundary condition” is not truly correct in inertial regimes within PDMS-made microchannels. It is not surprising that the results of numerical simulations must be validated with hard chips rather than soft (PDMS-made) microchannels. Apart from these issues, the inherent limitations in the standard microfabrication and soflithography techniquesmake researchers unable to explore particle migration in unconventional cross-sections (e.g., right-angled triangular or hexagonal). For instance, particle focusing within a triangular curved microchannel has never been explored due to the fabrication difculty. As such, there is a great need to develop standardized protocols for the fabrication of inertial microfuidic devices to facilitate ground-breaking research, while enabling quick translation into commercial products.

In this study, we have proposed a novel approach for the fabrication of inertial microfuidic devices based on the 3D printing method. Figure 1 demonstrates an overview of the fabrication process. Gaining the effici cy of a high-resolution 3D printer, the desired microchannel is printed while its face (where the design pattern exists) is outer, and the base is attached to the build plate (Fig. 1AI). Th s method is particularly signifi ant since the change of cross-sections in inertial microfuidics is of great interest. However, the printing parameters need to be optimized for the fabrication of a channel with proper and accurate dimensions. Te slice thickness

Figure 1. (A) Schematic illustration of the proposed workfl w for the fabrication of inertial microfuidic devices. I. Te desired channel geometry was printed by a high-resolution SLA/DLP 3D printer II. Afer cleaning the part by isopropanol, it was bonded to a PMMA sheet by means of a double-coated pressuresensitive adhesive tape. Te entire process just takes less than two hours. III. Beneftting from the PMMA transparency, high-speed, fuorescent, bright fi ld, or phase contrast microscopy can be performed from the bottom side of the channel (B) An actual complicated inertial microfuidic device containing a spiral and serpentine microchannel. (C) Fluorescent microscopy from the bottom side of the channel.

in Z direction, curing time of each layer, and total thickness of the part are the most critical factors to have a high-quality channel with a great surface fin sh. Various cross-sections, ranging from right-angled or isosceles triangular to hexagonal, were fabricated and the best optimized parameters were identifi d (Fig. S1). To complete the fuidic network, 3D-printed inertial microchannels need to be bonded to a substrate with enough optical transparency and rigidity for subsequent testing. In this work, a variety of scenarios has been evaluated, and upon extensive evaluations and characterizations, permanent bonding of 3D-printed channels to a PMMA sheet via a double-coated adhesive tape was selected as the most promising and reproducible method. A transparent double-coated pressure-sensitive adhesive tape (ARcare, Adhesive Research) having 25.4 µm clear polyester flm coated with AS-110 acrylic medical grade adhesive was cut with a similar size of PMMA sheet (Fig. 1AII). Afer the attachment of one side of the tape to the PMMA sheet, the 3D-printed inertial part was manually placed over the other side of the tape and pressed with a tweezer until no bubble was observed at the interface (Fig. S2).

An important feature of PDMS is its optical transparency, which makes it suitable for a broad range of microscopic applications. Given the fact that commercial DLP/SLA resins are not typically transparent, the attachment of 3D-printed microchannels to PMMA sheets provides enough transparency for the optical and fuorescent microscopy (Fig. 1AIII). What makes this approach attractive for a wide range of communities (e.g., biologists and chemists) is its user-friendliness for people without prior knowledge about microfabrication and soflithography. Te entire process from CAD drawing to printing and then testing takes less than 2 hours, portraying the versatility of this method for inertial microfuidic research. More importantly, devices made using this technique are not prone to the deformation and leakage compared to the PDMS-made devices, making them suitable to study new physics, especially at high Re. Furthermore, by considering the fabrication cost, time, and efforts of a complicated inertial microfuidic device, our suggested method is rapid and utilizes a low-cost raw material which are valuable features, especially in areas where resources are limited. Figure 1B,C depict a fi al device fabricated using this technique. Te internal channels are flled with red food color for the sake of illustration.

In order to investigate the bonding quality, a straight microchannel with dimensions of 50 µm height, 200 µm width, and 4 cm length was fabricated and tested accordingly. We have monitored the device performance for the appearance and growth of Safman-Taylor fi gers until it becomes stable, called “infation stability” (Fig. 2A). Te results are presented in a 2D diagram to identify the channel behavior at a given pressure, as shown in Fig. 2C. Our results revealed that the holding strength of double-coated adhesive tape was able to achieve a leak-proof interface between the 3D-printed part and PMMA sheet, not only at typical operating pressure reported in literature46, but also more than the capability of PDMS-made channels in withstanding high fl w rate conditions. Shear rate distribution across a line parallel to the channel width was also evaluated, and as Fig. 2B revealed, increasing the fl w rate leads to imposing more shear forces at the edges of the channels. Te more the fl w rate, the larger the appearance of Safman-Taylor fi gers (insets of Fig. 2B). Te green area in Fig. 2C shows the safe zone for performing inertial microfuidic experiments where no Safman-Taylor fi gers appear during the operation. We have found that at pressures more than 82.6 psi, Safman-Taylor fi - gers begin to appear; however, this does not impose any detrimental efect on the device performance (i.e., no leakage or bonding collapse). Also, we did not observe any delamination or deformation in channels afer consecutive runs at high pressures (i.e., 120 psi), all of which are common in PDMS-based inertial microfuidic devices (see Figs. S3 and S4 for the pressure drop, velocity profle inside the microchannels for a wide range of operating fl w rates).

The surface characteristic of the double-coated adhesive tape was also investigated using a proflometer. As Fig. 3 illustrates, the roughness of the tape is homogenous and is in the submicron range. Te values of Ra and Sa were about 250 and 240nm, respectively. Also, the roughness of the 3D printed parts was evaluated and value of Sa was less than 300 nm. Tese nanometric rugosities indicate that the roughness of tape does not have any efect on the fow profle and particle focusing. Although optically transparent, the optical characteristics of the PMMA sheet (2-mm-thick) and adhesive tape were evaluated to identify the possibility of accurate fuorescence imaging47. Hence, the UV-visible absorbance spectra for a wide range of wavelengths (i.e., from 200 to 1100nm) were recorded via a spectrophotometer (Cary 60 VU-Vis spectrophotometer, Agilent Technologies). Figure 3B,C reveal that the light loss is negligible for both PMMA and adhesive tape within the visible spectrum, resulting in no trace of autofuorescence residual.

Straight microchannel.
Straight microchannels with rectangular or square cross-sections are arguably the most widely used inertial microfuidic systems. Tanks to their ease of fabrication and the ability for parallelization, a myriad of applications have been developed using these platforms over the past decade48. Te required channel length for inertial particle migration to the equilibrium positions is L H = πμ /ρ α U f f m L 2 2 where fL is estimated in the range of 0.02 to 0.05 for (H/W) from 2 to 0.5, and the corresponding fl w rate for inertial migration is calculated as Q ≈ 2 / πμWH 3ρ αL fL 3 2 48. Channel Re (Re = ρUD/μ) and particle Reynolds number ( ) Re Re p H 2 = 2 α are two dimensionless numbers for the characterization of particle migration in a straight microchannel. When particle Re is much smaller than 1, viscous drag becomes dominant, and particles follow the streamline. Increasing particle Re augments inertial forces, causing inertial particle migration become obvious in the microchannel49,50.

Particle migration within a straight channel strictly depends on its cross-section. In square straight microchannels (with an aspect ratio (AR) (width/height) of 1), particles migrate to four equilibrium positions located at the center of each wall. Changing the cross-section to rectangular disturbs this focusing pattern where in a rectangular straight microchannel with AR of 0.5, focusing positions reduce to two near the center of long walls51. Th s behavior was explained by Zhou and Papautsky where they identifi d two-stage particle migration in rectangular straight microchannels21. Further increase in the AR results in the more unpredicted focusing behavior of particles. Generally, in channels with high AR, stable focusing positions are reduced. However, by exceeding Re from a critical value, the number of stable equilibrium positions increases which is a function of particle size, channel dimensions, and Re. Based on reported experimental results, = . κ κ ≤ ≤ ≤ ≤ − . Re A 697( R A / ) (4 5 / R R 60, 5 e 660) c 0 79 was identifed52. Te abovementioned results elucidate that particle focusing is strongly afected by channel cross-section. However, due to the fabrication limitations, dependency of various cross-sections to channel geometries was not systematically investigated. Recently, triangular and semi-circular cross-sections were fabricated using Si anisotropic etching with potassium hydroxide53, a brass for mold fabrication54, FDM for creation of sacrific al mold40, or unconventional micromilling14. However, critical fabrication limitations do not allow for further investigation on the dependency of triangular angle or type (e.g., right-angled triangular) on focusing patterns of the particles. Here, as a showcase, a straight microchannel with rectangular cross-section and AR of 4 (all channel dimensions are provided in Section S3 and Tables S1–5) was fabricated, and the results are illustrated in Fig. 4A. As the results indicate, for low Re (Fig. 4AI), 20µm particles focus at the center of long walls of the channel cross-sections, shown previously in PDMS-made microchannels. Nonetheless, the focusing pattern for particles at higher Re does not obey a specifc role. As clearly can be seen, increasing fl w rates leads to generation of additional focusing positions within the microchannels where side walls are also added to the equilibrium positions of particles (Fig. 4AII–IV). Furthermore, lateral migration of MDA-MB-231 and DU-145 cells at low fl w rates (10~20ml/hr) (Fig. 4BI–III) illustrates their single-line focusing within the rectangular straight microchannel, which is promising for fl w cytometry applications. Moreover, to showcase the versatility of the proposed method, a triangular straight microchannel was fabricated and the results are shown in Fig. 4D. Te results are completely in line with those reported in the literature where 10µm particles and cells occupy one lateral position in the channel for low fow rates (Fig. 4DI,II). For high

Figure 2. (A) Analyzing the Safman-Taylor fi ger criteria for the bonding quality in a microchannel versus various fl w rates. For fl w rates lower than 1.5ml/min, the Safman-Taylor fi gers do not appear, while for fl w rates more than 1.5ml/min, Safman-Taylor fi gers become discernible. (B) Shear rate distribution across a line parallel to the channel width. (C) Te more the pressure, the faster the creation of Safman-Taylor fi gers. In the green area, Safman-Taylor fi gers do not appear during the experiments. Also, the results show that there is not any evidence of channel burst or delamination during the bonding quality test.

Figure 3. (A) Surface topography of the double-coated adhesive tape. Results in a vertical line (green line), horizontal line (blue line), or across an area (red rectangular) show that the tape has homogenous roughness with a nanometric value, which does not impose any interference on the channel performance. Te absorbance amount of (B) double-coated adhesive tape and (C) PMMA sheet, implying that these two materials are transparent for visible spectra range, and there is not any signifi ant absorption.

Sinusoidal and serpentine microchannel.
Inertial microfuidics in sinusoidal (curvilinear) microchannel has gained traction due to its improved focusing performance, the ability of massive parallelization, and small footprint. In the sinusoidal microchannel, the curvature direction changes in each loop, resulting in an intricate phenomenon that help in particle focusing. Indeed, by alternating the curvatures, the direction of Dean fl ws changes, and secondary fl ws may not reach a steady-state condition. Th s design was fi stly developed by Di Carlo in 2007 and its capability in wide ranges of Dean number ( / De = Re D R2 h , where Re is channel Re, Dh is characteristic length of channel, and R is the radius of channel curvature) was evaluated56. By assuming that Dean drag forces were balanced with shear gradient lift forces, his team proposed the ratio of inertial lift forces to Dean drag forces as F F L D / 2 = ra /Dh 2 3 (where r is radius of channel and a is particle diameter). Generally, if FL/FD ≫ 1, secondary fl ws do not afect particles, and if FL/FD ≪ 1, particles are entirely afected by secondary fl ws57. Th application of this microchannel was even expanded where it was used for high-throughput separation of micron and sub-micron bioparticles (cyanobacteria)58 and a microfuidic concentrator for harvesting of cyanobacteria59. In addition, in a comprehensive study, the design principle of curvilinear microchannels was investigated, and a map for various focusing phenomena was provided, based on F F/ ( ~ Re / ) De ( / a D ) f L D h L 2 2 3 where fL was approximated by Zhou and Papautsky21 as fL ~ 1/Re(Dh/a)2 60. Te dependency of curvature angle61 and various cell lines62 on the focusing positions was also evaluated. In order to showcase the adaptability of our method for fabrication of various inertial microfuidic devices, a curvilinear microchannel with rectangular cross-section (Fig. 5A) was designed, fabricated, and evaluated. Figure 5B reveals that 15 µm particles fi st occupied two focusing positions and by increasing the fl w rate, it reduced to a single focusing line across the channel (e.g., at 5th loop for fl w rate of 900µl/min (Fig. 5C)) which is consistent with the previously reported results60.

By altering the design of curvilinear to straight, serpentine microchannel with a square-wave pattern is created. Th s channel proves to have unique features for size-based particle focusing. Generally, particle focusing achieves when a/Dh > 0.07 and Rp~1. However, additional secondary fl ws in serpentine microchannel lead to focusing of particles with smaller diameters compared to those calculated by the above formula. Tese devices can beneft from parallelization along their vertical direction, thereby increasing their throughput. Tree focusing

Figure 4. (A) Inertial microfuidics in a rectangular straight microchannel with height and width of 50 µm and 200 µm, respectively. I. At fi st, 20 µm particles occupy the center of the channel as their focusing position. Te intensity profle also illustrates that particles are focused at the center of the channel. II–IV. Later, looking at lateral position and intensity profles reveal that by increasing the fl w rate, side walls are added to the focusing position of the particles, and the focusing band of particles at center becomes wider. To extract these images, we have used “max intensity” feature from Fiji Sofware (https://fji.sc). (B) Te equilibrium position of MDA-MB-231 cells at fl w rates of I. 10ml/hr and II. 20ml/hr and III. DU-145 cells at fl w rate of 20ml/hr confi ms the single-line focusing of cells within the rectangular straight microchannel (from top view). (C) A surface proflometry of the rectangular cross-section which shows the rectangular profle of the microchannel. Results show that the channel has perfect shape and quality which is suitable for inertial microfuidics. (D) In triangular microchannel, particles fi st migrate to I. and II. one focusing position and then this increases to III. three separate points. Th s trend is similar to those reported in the literature14,53,54. Te results for MDA-MB-231 cells at fl w rate of 200 µl/min illustrate a single-line focusing position, and at fl w rate of 500 µl/min depict three focusing positions. (E) Surface proflometry of the triangular straight microchannel with height and width of 40 µm and 300 µm, respectively

Figure 5. (A) Inertial microfuidics in a curvilinear microchannel with height and width of 50 and 200µm, respectively. (B) Te results show that the equilibrium position of particles depends on the fl w rate and has diferent focusing modes. Particles fi st focus at two equilibrium positions and then occupy just one focusing line. Eventually, by further increasing the fl w rate, Dean drag forces become dominant, resulting in defocusing of particles. Te trend is similar to that reported in the literature60. (C) 15µm particle migration throughout the channel for fl w rate of 900µl/min, demonstrating that particles are focused at the 5th loop. (D) Inertial microfuidics in a serpentine microchannel with height and width of 40 and 200µm, respectively. Te number of lateral positions depends on the applied fl w rate at the entrance of the channel. (E) Focusing behavior of 10µm particles at 0.7ml/min. (F) As intensity profle elucidates, at the 10th loop, particles reach the stable equilibrium position. Lateral migration of MDA-MB-231 cells at fl w rate of (G) 0.7ml/min and (H) 0.8ml/min shows a single-line focusing of these cells at the center of the channel. (I) Surface proflometry of the channel with rectangular cross-section with width and height of 200 and 50µm, which shows the accuracy and high-quality of the channels appropriate for inertial microfuidics.

patterns can be identifi d by increasing the input fl w rate, i.e., two-sided focusing, transition focusing, and central single-line focusing. If inertial efects dominate the secondary fl ws, particles occupy two lines near the walls. In contrast, dominance of secondary fl ws results in a single-line focusing at channel center. If these two efects have the same order, particles focus as a wide streak. Gaining the efciency of two-sided focusing for small particles and central focusing for big ones give us the opportunity of size-based particle separation. Based on the literature, a serpentine microchannel with cross-section of 40×200µm (H×W) and 15 loops was fabricated and used to showcase the focusing of 10 µm particles (Fig. 5D). In a straight channel with 40×200 µm (H×W) cross-section, a/Dh for 10 µm is 0.056, which is less than the focusing criteria (0.07); theoreically, these particles cannot focus in a straight microchannel. However, in the serpentine microchannel with the aid of secondary fl ws, 10 µm particles can effici tly be focused. Figure 5E elucidates that 10 µm particles at fow rate of 0.7ml/min can be focused at the center of channel at the 10th loop and occupy central equilibrium position at the outlet, which is consistent with the insets as normalized intensity profle of the particles (Fig. 5F). More importantly, the performance of device was tested with MDA-MB-231 cells (Fig. 5G,H), and the results are consistent with those reported in the literature63. Te surface proflometry of the channel cross-section is also provided in Fig. 5I, indicating the high-accuracy of the proposed method for fabrication of inertial microfuidic devices.

Spiral microchannel.
Spiral defi es as a curve winding around a center point with continuous decreasing or increasing manner. When fl w passes the curvature, velocity mismatch occurs in the curve section of the channel, resulting in the generation of secondary fl ws. In inertial microfuidics, spiral microchannel has progressed signifcantly, and nowadays, most of the particle/cell separations are performed using these microchannels64. De is used for the characterization of secondary fows within the channel. Intuitively, smaller channel curvature or larger channel size or Re leads to higher De, thereby imposing stronger secondary fl ws within the channel. For a given De, average transverse Dean velocity (UDe=1.84×10−4 De1.63) and Dean drag force = = πµ α π . × µ α − . ( 3 F U 5 4 10 De ) D De 4 1 63 can be identifi d. However, the exact behavior of particle migration at the downstream of the fuid was not thoroughly investigated, and all results are based on experimental data. Te most appealing feature of spiral inertial microfuidics is its high-throughput where 2100 particles per second can be processed9 . Particle sorting is one of the most signifcant applications of spiral microfuidics. Previously, the potential of a rectangular spiral microchannel for continuous and simultaneous isolation of 10, 15, and 20µm based on soflithography was investigated (Fig. 6AI) 65. Dean fl w dynamics for a low-aspect-ratio rectangular spiral microchannel was also thoroughly explored66. Beyond a simple rectangular spiral microchannel, various geometry modifcations for regulation of Dean forces and performance enhancement of the device have been proposed. Beneftting from micromilling (Fig. 6AII), trapezoidal spiral microchannels illustrate promising results in redistribution of lateral focusing positions of particles appropriate for size-based particle separation. In these channels, smaller particles focus along the outer wall, whereas larger ones migrate toward the inner wall67. Th s superior advantage has been widely investigated by our group, among other groups, for circulating tumor cell (CTC) and circulating fetal trophoblasts (CFT) isolation19,68, blood plasma separation69, isolation of microcarriers from mesenchymal stem cells70,71, microalgae separation72, and synchronizing C. elegans73. Also, multiplexing using stack of attached PDMS layers to boost the throughput is illustrated previously69,74. However, most of the aforementioned applications are just doable by utilizing cleanroom facilities or employing conventional micromachining (e.g., metal machining or laser cutting) for the fabrication of microchannel. Besides, micromachining has its own limitations such as inability to make sharp corners or difculty in making spiral loops close to each other. Tese challenges highlight an unmet need for the fabrication of spiral microchannels using a versatile method which is robust and can surmount aforementioned issues.

As a showcase of the versatility of our proposed method, we have fabricated a spiral microchannel with trapezoidal cross-section with a width of 600 µm and heights of 80 and 130 µm. Tese results are then put aside a PDMS chip with similar dimensions, and the data is provided in ESI (Fig. S5). Despite all progress in spiral inertial microfuidics, there is not any report of a spiral with cross-sections rather than rectangular or trapezoidal. In other words, a huge amount of potential remains intact to study spiral microchannels with diferent cross-sections such as triangular (Fig. 6AIII). For this aim, for the fi st time, we have fabricated a spiral microchannel with right-angled triangular cross-section (as schematically shown in Fig. 6B) where the width and height are 600 and 210µm, respectively. As the results are illustrated in Fig. 6C, there is a tight band focusing for particles larger than 10 µm, which is suitable for high throughput fl w cytometry applications where single line focusing is desired. Also, we observed double-band focusing behavior for 20µm particles at fl w rate of ≥4ml/min. Te dimensions (Fig. 6D) and channel cross-section (Fig. 6E) show the accuracy of the proposed method the for fabrication of right-angled triangular spiral microchannel (check Fig. S6 for contraction-expansion array microchannel results). Tese results illustrate the flex bility of this method where a complex cross-section can be fabricated in less than two hours with high robustness and stability. Our results hold promise for leveraging the potential of additive manufacturing for the fabrication of inertial microfuidic devices, which is more challenging using conventional microfabrication methods (see Section S6 for multiplexing of 3D printed inertial microfuidic devices).

Cellular studies.
PDMS-made inertial microfuidic devices have been widely used for the cell separation using biological samples such as blood and urine. While PDMS is proven to be a biocompatible material with minimum side efects on cells, we have tested the 3D printed devices using DU145 cells, assessing their viability and functionality post-separation. Te collected cells from the device outlet were cultured back into a petri dish for 5 days, showing similar morphological features to the control group as shown in Fig. 7A,B. Te fl w cytometry tests (Fig. 7C) showed that the viability of the cells was not compromised during the operation using 3D printed devices. Te real-time PCR analysis was utilised to assess the expression of genes related to the general activities and stress responses in both treated and untreated cells (Fig. 7D). Te similar expression level of GAPDH and CDKN2A confi med that neither cellular metabolism nor cell cycle progression were afected afer processing

Figure 6. (A) Illustration of a spiral microchannel where the fuid direction is from outside to inside. I. Firstly, several groups (e.g., Bhagat et al. 9 , Papautsky et al. 66,77, etc.) have shown the capability of rectangular spiral microfuidics for such applications as fl w cytometry or microparticle/cell separation. Te fabrication of these devices was based on photolithography. II. Gaining the effici cy of micromilling, many groups (e.g., Guan et al. 67, Warkiani et al. 20, etc.) made an attempt to get the advantage of trapezoidal spiral microchannel for particle/cell fltration and fractionation. III. In this study, for the fi st time, we have shown the fabrication of a right-angled triangular spiral microchannel with the aid of additive manufacturing. (B) Schematic illustration of the microchannel where the inset shows the cross-section of the microchannel. (C) Results reveal that for particles larger than 10 µm, a tight focusing band appears at the outlet of the channel. Also, for larger particles at high fl w rates (i.e., 4 ml/min) double-band focusing appears. (D) dimensions of the right-angled triangular spiral microchannel where the inner wall is 210 µm and the width is 600 µm. Te hydraulic diameter of this channel is similar to a spiral with a trapezoidal cross-section and the dimension of 80×130×600 µm. (E) Illustration of a right-angled triangular cross-section, which shows the accuracy of the fabrication process.

through the microchannels. Also, there are no signifi ant changes in the expression of TXNIP and MAPK14, which are two well-known regulators of cellular stress. In all, the 3D-printed inertial microfuidic device does not alter the cell activity and is safe to be used in biological assays.

Conclusion
In this study, we have showcased a robust and versatile workfl w for the manufacturing of inertial microfuidic devices for both laboratory experiments and industrial applications. Te proposed method involves 3D-direct printing of a channel with the desired structure and geometry via a high-resolution DLP/SLA 3D printer. Our approach relies on the bonding of 3D-printed devices (i.e., with high-quality surface fin sh) to a transparent PMMA sheet via a double-coated pressure-sensitive adhesive tape. Given the great transparency of PMMA layer, these devices can be utilised in bright fi ld, phase contrast, fuorescent, or high-speed microscopy. Te bonding quality was evaluated by Safman-Taylor fi ger criterion, and the results showed that the device is capable of withstanding pressure as high as 150psi (nearly triple of the value reported for PDMS). Te versatility of this method allowed us to fabricate and evaluate a plethora of existing inertial microfuidic devices as exemplifi d for fabrication of straight, spiral, serpentine, curvilinear, and contraction-expansion arrays microchannels. Te total time frame, from designing a part to starting an experiment, takes less than two hours, allowing multiple experiments in a single day. Also, for the fi st time, we have fabricated and examined a new inertial microfuidic device, i.e., spiral microchannel with right-angled triangular cross-section which is theoretically impossible to fabricate using photolithography. We believe that the proposed workfl w will provide inertial microfuidic devices within the reach of any research groups involving in particle/cell manipulation without strong microfabrication background.

Figure 7. (A) Monitoring the morphological feature of cells during fve days post-experiment, compared to the control group. (B) Fluorescent staining of F-actin flaments in expanded cells on day 5 (green = phalloidinFITC, blue = nucleus) (C) Representative plot and mean value ± SEM of fl w cytometric analysis of live/dead population for control and test group. (D) Cycle of threshold (Ct) value (expression level) for GAPDH (cellular metabolism related gene), CDKN2A (cell cycle regulatory gene), TXNIP, and MAPK14 (genes involved in cellular stress response). Results are expressed as mean value ± SEM from three independent experiments. Scale bars are equal to 100 µm in large images and 20 µm in inset ones

Materials and methods

Fabrication method
In this study, inertial microfuidic devices were fabricated using a high-resolution DLP/SLA 3D printer (Ultra 50, Miicraft, Hsinchu, Taiwan) featuring 30 µm XY resolution and 32×57×120 mm3 printing area. Te desired inertial microfuidic device is fi st drafed using a commercial CAD drawing sofware (SolidWorks 2016) and then translated into STL format, a suitable fle for 3D printer language. Te fle is then sliced in Z direction using the Miicraft sofware (Version 4.01, Miicraft). Te slicing in Z direction (slice thickness) can be adjusted from 5 to 200µm with an increment of 5µm. Te slicing option is related to the complexity of the geometry. In geometries with ramp or step, the slice thickness should be small, whereas for planar or orthogonal structures, higher slice thickness is suitable. Te detailed dependencies of printing parameters are provided in electronic supplementary information (ESI) (Section S1 and Fig. S1). Te sliced fle is then sent to the 3D printer with UV wavelength of 358–405nm. Te UV light is projected from the bottom of the resin bath (flled with BV-007 resin) and passes through a transparent Tefl n flm. BV-007 is an acrylate-based resin containing 80–95% acrylate components and 10–15% photoinitiator and additives. Once the UV light cures one layer, the Z-stepper motor moves one slice upward, and the next layer starts to be polymerized. Th s process continues until the part is printed, successfully. When the part is removed from the picker, it should be rinsed and washed with isopropanol, thoroughly and air-dried by an air nozzle. Aferward, the microchannel is exposed to a UV light with 405 ± 5nm wavelength within a curing chamber for post curing process. One superior advantage of this method compared to the soflithography is that it does not require punching the holes for inlets and outlets since those are printed as a one-body part by the 3D printer. Eventually, tubes (Tygon tubing, inner diameter: 0.020″, outer diameter: 0.060″) were connected to the microchannel by a tweezer.

Preparation of bead suspension
Fluorescent microbeads (Fluoresbrite Microspheres, Polysciences Inc, Singapore) with 0.01% volume fraction and various diameters were added to the MACS bufer. Te primary usage of MACS bufer is to prevent nonspecific adhesion of microbeads to the tubing of the microchannel. Te distribution of particles is illustrated using standard deviation, minimum, or maximum light intensity plots, as reported previously54. Preparation of bead suspension Fluorescent microbeads (Fluoresbrite Microspheres, Polysciences Inc, Singapore) with 0.01% volume fraction and various diameters were added to the MACS bufer. Te primary usage of MACS bufer is to prevent nonspecific adhesion of microbeads to the tubing of the microchannel. Te distribution of particles is illustrated using standard deviation, minimum, or maximum light intensity plots, as reported previously54.

Bonding quality test
Inertial devices are operated at high fl w rates; hence, the bonding technique must provide enough strength to prevent leakage from the interface. To evaluate the bonding quality of our proposed technique, a simple 3D-printed straight channel featuring 50 µm height, 200 µm width, and 4 cm length was bonded to a 2-mm-thick PMMA layer. A high-pressure syringe pump (Chemyx Fusion 4000, Chemyx, TX, USA) was used to inject fuids inside the channel from a small syringe (6ml). Increasing the fl w rate leads to the generation of Safman-Taylor fi gers around the inlet, in which the most pressure in the channel present (Section S2). Safman-Taylor fi gers are generated by the movement of a viscous fuid within a porous material75,76. As the bonded adhesive tape forms a porous zone between the connecting parts, this theory is applicable for the bonding evaluation. An increase in the applied pressure leads to developments of the Safman-Taylor fi gers until the bonding fails. A CCD camera (DP80, Olympus, Tokyo, Japan) mounted on an inverted microscope (IX73, Olympus, Tokyo, Japan) was used for monitoring the bonding integrity. All recorded data were obtained immediately afer the bonding of the 3D-printed channels to a PMMA sheet.

Surface characterization
For the surface characterization of double-coated adhesive tape, a 3D laser microscope (Olympus LEXT OLS5000) was used, and an LMPLFLN 20x LEXT objective lens (Olympus) was selected. Arithmetic mean deviation (Ra), the arithmetic mean of absolute ordinate Z (x, y) documented across a line, and arithmetical mean height (Sa), the arithmetic mean of the absolute ordinate Z (x, y) recorded across a region were chosen to evaluate the surface characterization of the tape.

Cell culture, harvesting, and device operation
DU145 cells (human prostate cancer cell line) were cultured and expanded under standard culture condition (37 °C and 5% CO2) using Roswell Park Memorial Institute medium (RPMI, TermoFisher) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% Penicillin Streptomycin (Pen/Strp, Gibco). Cells were harvested when the fask was 80% confuence. To obtain a homogenous cell suspension without cell clumps, a suffici t volume of TryplE (Gibco) was added to cover the whole fask, and the fask was incubated at 37 °C for 5 min. Te cells were then collected in a 15 ml falcon tube and counted with a hemocytometer. TryplE was then replaced with phosphate bufer saline (PBS, Gibco), and cells were diluted to 106 cells/ml concentration by PBS. Aferward, cells were introduced to 3D-printed microchannels (straight rectangular microchannel with a length of 4 cm, width of 200µm, and height of 50µm) at the fl w rate of 0.3ml/min. A group of untested cells was kept as control.

Morphological analysis and cell viability assay
In order to evaluate the viability of cells afer passing through the channels, a live-dead assay was performed using Live and Dead Cell Assay kit (Abcam, Cambridge, UK) afer the test. One group of collected DU145 cell suspension was diluted to 5×105 cell/ml and incubated with the staining solution for 10min under room temperature, based on the kit manufacturer’s instruction. Ten, the stained cells were processed through fl w cytometry (Olympus CKX53, Tokyo, Japan) and analyzed by CytExpert sofware (Beckman Coulter, Inc.). Live and dead cells were detected by green (λ excite/emit = 488/515nm) and red (λ excite/emit = 488/617nm) fuorescent, respectively. Another group of DU145 cells was cultured in a culture fask and stained by live and dead staining afer 24 hours without detaching by TryplE. Fluorescent microscopic imaging was taken for adhered cells to identify live cells (with green cytoplasm) from dead ones (with red nucleus) (Fig. S8). Te morphology of the attached cells was monitored under an inverted microscope for up to fve days. Furthermore, the morphological features of both test and control groups were visualized by fuorescence staining of cytoskeleton afer three days. Te F-actin flaments were fi ed and permeabilized by 4% paraformaldehyde (PFA, Sigma) and 0.2% Triton X-100 (Sigma) and then labeled by Phalloidin- FITC (Sigma).

Real-time PCR analysis
The effect of shear stress on the expression of genes related to proliferation and survival was measured by real-time PCR (BioRad CFX 96 thermocycler). Briefy, the processed cells were reseeded on a culture dish and kept under standard culture conditions for one day. Next, total RNA of cells were extracted by using PureLink RNA Mini Kit (TermoFisher), and cDNA was synthesized by applying Revert Aid First Strand cDNA Synthesis Kit (TermoFisher). Real-time PCR was performed using specific TaqMan primer sets and TaqMan PCR Master Mix (TermoFisher) with following cyclic conditions: 95 °C for 10min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 1min, and 72 °C for 10 s.

Ultrasensitive and rapid quantification of rare tumorigenic stem cells in hPSC-derived cardiomyocyte populations

by

Ultrasensitive and rapid quantification of rare tumorigenic stem cells in hPSC-derived cardiomyocyte populations

Zongjie Wang 1 2, Mark Gagliardi 3 4, Reza M Mohamadi 5, Sharif U Ahmed 5, Mahmoud Labib 5, Libing Zhang 5, Sandra Popescu 5, Yuxiao Zhou 5, Edward H Sargent 1, Gordon M Keller 3 4 6, Shana O Kelley 2 5 7

The ability to detect rare human pluripotent stem cells (hPSCs) in differentiated populations is critical for safeguarding the clinical translation of cell therapy, as these undifferentiated cells have the capacity to form teratomas in vivo. The detection of hPSCs must be performed using an approach compatible with traceable manufacturing of therapeutic cell products. Here, we report a novel microfluidic approach, stem cell quantitative cytometry (SCQC), for the quantification of rare hPSCs in hPSC-derived cardiomyocyte (CM) populations. This approach enables the ultrasensitive capture, profiling, and enumeration of trace levels of hPSCs labeled with magnetic nanoparticles in a low-cost, manufacturable microfluidic chip. We deploy SCQC to assess the tumorigenic risk of hPSC-derived CM populations in vivo. In addition, we isolate rare hPSCs from the differentiated populations using SCQC and characterize their pluripotency.

We kindly thank the researchers at University of Toronto for this collaboration, and for sharing the results obtained with their system.

Introduction

Human pluripotent stem cells (hPSCs) hold great promise for cell therapy given their ability to differentiate into many different cell types (1). Numerous studies have demonstrated the considerable potential of hPSC derivatives in treating chronic diseases, including neuron degeneration (2) and chronic heart failure (3). Typically, a dose of cell therapy for treating heart failure requires 0.1 to 1 billion of de novo hPSC-derived functional cells (4); therefore, the translation of cell therapy from bench to bedside heavily depends on reliable manufacturing of high-quality cell products (4–6).

The self-renewal and pluripotent properties of hPSCs are also associated with a high level of tumorigenicity in vivo (7). Undifferentiated cells can persist in differentiated populations following long periods of time in culture (8, 9), and rare contaminating hPSCs, even at concentrations less than 0.025%, can lead to teratoma formation in animal models (10–12). As a result, quantitation of the percentage of rare hPSCs is a key quality control parameter that needs to be monitored in manufactured populations to be used for cell therapy applications (4, 13).

Flow cytometry (FCM) and the polymerase chain reaction (PCR) are powerful methods for the analysis of rare cells. Unfortunately, neither of these methods is sensitive enough to rapidly and accurately identify rare hPSCs in relevant samples. FCM has intrinsic limitations including sampling losses and dead volumes (14) that reduce accuracy at exceptionally low levels of hPSC contamination that may still represent a potential risk of tumor formation (15). PCR-based methods are often problematic due to the high background from differentiated cells relative to the rare undifferentiated controls (16) and reverse transcription–induced artifacts, including primer-independent complementary DNA (cDNA) synthesis and template switching (17).

Here, we report a new approach to the quantitation of rare undifferentiated cells: stem cell quantitative cytometry (SCQC). This method takes advantage of a rare cell profiling approach based on a strategy for monitoring cancer cells (18, 19). SCQC uses a microfluidic chip that is scalable, cost-effective, and compatible with the requirements of manufacturing and quality control. SCQC has excellent sensitivity and is able to profile rare hPSCs robustly even when present at concentrations as low as 0.0005% in populations of hPSC-derived cardiomyocytes (CMs). Through the analysis of CM samples containing defined numbers of spiked hPSCs, we demonstrate that SCQC detects rare contaminants with unprecedented performance. Comparative studies show that SCQC can accurately quantify hPSCs at levels that are not reliably detected by either FCM or droplet digital PCR (ddPCR). Last, we use SCQC to isolate live rare hPSCs from differentiated CM populations and characterize their pluripotency.

Materials

A Bottle of the Master Mold for PDMS Device Resin

Master Mold Resin

Learn More

H Series

Learn More

Results

Development of microfluidic SCQC
The device used for SCQC relies on immunomagnetic labeling to profile cells based on their surface marker expression (Fig. 1A) (20). Cells are labeled with magnetic nanoparticles (MNPs) that bind to a surface marker expressed by hPSCs, but not CMs. Labeled cells are subsequently introduced into a microfluidic device with a flow velocity gradient and a constant magnetic field (Fig. 1, A to C, and fig. S1, A and B). Cells with more MNPs experience a higher magnetic force to compensate the downstream drag force generated by the flow velocity. As a result, the hPSCs that bind more MNPs than the differentiated CMs can withstand high flow velocities and remain in the high-velocity regions of the device. The CMs with few or no MNPs are captured in the low-velocity regions or flushed away. At the completion of the run, the cells trapped in the microfluidic device are quantified by immunofluorescence and microscopy to generate a cytometric profile that includes number, phenotype, and distribution of trapped cells. The information from the cytometric profile provides an assessment of the total number of hPSCs.

Fig. 1 SCQC for rare stem cell analysis.
(A) Overview of the SCQC approach. Cells are labeled with MNPs functionalized with an antibody against specific stem cell surface markers. Labeled cells are magnetically captured in a 3D-printed microfluidic device with a flow velocity gradient. Stem cells labeled by a high number of MNPs can withstand higher flow velocities and are captured in earlier zones. The number of stem cells in each zone is quantified by immunostaining and microscopy to generate a cytometric profile that can be used for further analysis. (B) Illustration of the microfluidic device for SCQC. Eight sequential capture zones with a decreasing flow velocity gradient are generated by linearly increasing the channel height. (C) Picture of the fabricated device. A dye-containing solution was introduced into the channel to visualize the changes in channel height. Error bar indicates the SD of the mean from three experiments (B).

Sensitivity and specificity of SCQC

We used the HES2 hPSC line and derivative CMs to characterize and optimize the performance of SCQC. The CMs were differentiated for at least 20 days following an established protocol (fig. S2) (21, 22). The hPSC-derived CMs contain more than 85% cardiac troponin T (cTNT)–positive cells. In the first suite of experiments, we used FCM to benchmark eight surface markers (SSEA-1, SSEA-4, TRA-1-60, TRA-1-81, EpCAM, CD90, CD9, and E-cadherin) and three intracellular markers (Oct4, SOX2, and Nanog) for distinguishing hPSCs and hPSC-derived CMs (fig. S3). FCM results revealed that TRA-1-60 and EpCAM offer the highest separation (equivalently the best contrast) among all markers. Hence, we chose TRA-1-60 and EpCAM MNPs for on-chip optimization.

HES2 hPSCs or derivative CMs were labeled with antibody-conjugated MNPs, profiled using SCQC and stained by a (4′,6-diamidino-2-phenylindole) DAPI/NucDead 488 cocktail for visualization (fig. S4A). TRA-1-60 MNPs produced higher levels of hPSC capture and hPSC-derived CM depletion (fig. S4B) than EpCAM MNPs. We also investigated the specificity of the TRA-1-60 MNPs for labeling hPSCs by transmission electron microscopy (fig. S4C). More than 10 major clusters of MNPs were observed on the surface of hPSCs, while 0 or 1 major cluster of MNPs was detected on the CMs. On the basis of these findings, we concluded that the TRA-1-60 has enhanced performance for capturing rare hPSCs in hPSC-derived CMs compared with other surface markers. Next, we tested the direct isolation of hPSCs using TRA-1-60 MNPs via magnetic-activated cell sorting (MACS). However, MACS had poor capture and recovery efficiency for both live and fixed cells (fig. S5). Notably, around 20% of cells, regardless of cell type, remained trapped in the columns and could not be recovered. This is consistent with previous studies (23, 24). Therefore, it is necessary to use the microfluidic chip to enable more efficient recovery and in situ immunostaining of rare cells for quantification.

We next assessed the specificity of SCQC to evaluate its utility for the quantitation of undifferentiated hPSCs. In the absence of an external magnetic field or the absence of MNPs, less than 0.05% hPSCs were captured at the flow rate of 2 ml/hour. Subsequently, the flow rate, which dominates the flow velocity on-chip, was optimized to balance the capture of hPSCs and the depletion of hPSC-derived CMs (fig. S4D). At the optimized flow rate (10 ml/hour), we captured 85.7% of hPSCs while depleting 99.7% of CMs (Fig. 2, A and B). Over 90% of captured hPSCs were detected in zones 1 to 5, which generates a reproducible cytometric profile. In addition, we characterized the dynamic range of the SCQC using the optimized flow rate. The chip had a consistent capture efficiency and cytometric profile in the range of 0 to 5000 hPSCs (Fig. 2C and fig. S4E). The chip offered sufficient depletion (>99%) up to 500,000 CMs (fig. S4E). Also, the chip maintained a consistent capture performance when using other hPSC cell lines, including an induced pluripotent stem cell line (fig. S4F) and other hPSC-derived populations (i.e., definitive endodermal cells) (fig. S4G). On the basis of these results, we conclude that SCQC is a highly sensitive and selective method with an excellent dynamic range for the capture and analysis of hPSCs.

Fig. 2 Characterization of sensitivity and selectivity of the SCQC approach. (A and B) Representative cytometric profiles of captured (A) hPSCs and (B) hPSC-derived CMs at the flow rate of 10 ml/hour (n = 3). (C) Capture performance of hPSCs in CM-free buffer (n = 3). (D) Representative microscope images of captured hPSCs in CMs. HPSCs are quantified as DAPI+, Oct4+, and Nanog+. (E) Capture performance of hPSCs spiked in 1 million hPSC-derived CMs (n = 4). Error bar indicates the SD of the mean from all experiments (A to E). Cell capture experiments (A, C, and E) were performed at the flow rate of 10 ml/hour and the volume of 1 ml. The number of hPSCs (A) or CMs (B) was 500.

We determined the limit of detection (LOD) of SCQC using samples of hPSC-derived CMs spiked with defined numbers of hPSCs. As the SCQC device nonspecifically captures a small fraction of hPSC-derived CMs, immunostaining was used to quantify the number of hPSCs on the fluidic chip. The hPSCs were defined using a cocktail of DAPI, Oct4, and Nanog (Fig. 2D). We found that SCQC can clearly identify the difference between the negative control (zero hPSC in 1,000,000 hPSC-derived CMs) and the 0.0005% sample (five hPSCs in 1,000,000 hPSC-derived CMs as shown in Fig. 2E). Hence SCQC achieves a LOD of 0.0005% for quantifying rare hPSCs.

Quantitative comparison between SCQC, FCM, and ddPCR

We conducted a comparative study to systematically evaluate the performance of SCQC, FCM, and ddPCR for rare hPSC detection. We generated populations of hPSC-derived CMs containing 0.01 to 5% of spiked HES2 hPSCs. For FCM, we used TRA-1-60 and EpCAM as the hPSC markers with a two-laser six-color flow cytometer. For ddPCR, we monitored the expression of three hPSC genes: POU5F1, SOX2, and CD326. TBP or B2M was included as a housekeeping control. For SCQC, we applied the TRA-1-60 MNPs and the flow rate of 10 ml/hour, as detailed above.

The representative profiles obtained by FCM are shown in Fig. 3A and fig. S6A. The hPSC signal (TRA-1-60+ EpCAM+) decreased rapidly with the decreasing number of hPSCs. FCM was unable to detect hPSCs in the 0.1% sample, as the signal for these samples was the same as the signal for the samples that did not contain hPSCs. The combination of the two most sensitive markers (TRA-1-60/EpCAM) yielded the optimal LOD around 0.2% to 0.3% (see fig. S6, A and B). Previous literature suggests that the number of total events that must be collected to detect a population present at a frequency of 0.1% is at least 10 million (25). Routine collection of 10 million events by FCM is neither cost- or time-effective.

Fig. 3 Benchmarking the performance of SCQC, FCM, and ddPCR for rare hPSC quantification. (A) Representative cytometric profile of FCM for samples with spiked hPSCs. (B and C) Detection performance of ddPCR using EpCAM and (B) TBP or (C) B2M as the housekeeping control for normalization. (D to F) Detection performance of SCQC and FCM in the range of (D) 0 to 5%, (E) 0 to 2%, and (F) 0 to 0.1% (n = 3 for SCQC, FCM, and ddPCR; 50,000 cells were analyzed for each replicate). Error bar indicates the SD of the mean from three experiments (B to F). Cell capture experiments (D to F) were performed at the flow rate of 10 ml/hour using a total volume of 1 ml. Each cell suspension contained 50,000 hPSC-derived CMs spiked with various amounts of undifferentiated hPSCs in the desired final concentration, as indicated on the x axis.

The representative ddPCR results are shown in Fig. 3, (B and C) and fig. S6 (C and D). From the three primer sets tested (POU5F1, SOX2, and EpCAM), we found that the combination of EpCAM as the target and TBP as a housekeeping gene yielded the lowest LOD, calculated at 0.2 to 0.4% (Fig. 3B). The LOD obtained using POU5F1 and SOX2, which are the definitive genes for pluripotency, was higher than 1% (fig. S6D). This is not an unexpected result as previous literature has indicated that pluripotent genes (i.e., POU5F1, SOX2, Nanog, Klf-4, and Lin28) are weakly expressed at the RNA level in hPSC-derived cells, typically in the range of 0.01 to 1% relative to undifferentiated hPSCs (16, 26). As a result, the signal from rare hPSCs is concealed by the large background generated by a substantial excess of hPSC-derived cells. Therefore, given the lack of a specific hPSC marker, ddPCR is not a practical method for quantifying rare hPSCs in differentiated population.

We summarized the comparative results in Fig. 3 (D to F). The LOD of SCQC, FCM, and ddPCR is <0.01, 0.2 to 0.3, and 0.2 to 0.4%, respectively (table S1A). In addition to having a superior LOD, SCQC also offered the best linearity of detection (R2 > 0.99) among the three technologies (table S1B). On the basis of this comparative study, it is clear that SCQC outperforms the existing cell detection methods when quantifying rare hPSCs in batches of differentiated CMs. In addition, SCQC also has other advantages for cell manufacturing including cost-effectiveness, scalability, and compatibility with the U.S. Food and Drug Administration Current Good Manufacturing Practice (cGMP) regulations (27) (table S2). SCQC facilitates accurate detection and quantitation of undifferentiated cells within an hour at a cost of $30 per chip per run. Given its compact design and scaled fabrication method, this technology can be easily set up to support parallelized and large-scale operations.

Analysis of the tumorigenicity of rare hPSCs

To benchmark whether the performance of SCQC is relevant for the analysis of batches of therapeutic cells, we used SCQC to assess the tumorigenicity of samples containing different levels of hPSCs (Fig. 4). We prepared samples of hPSC-derived CMs (1 million) spiked with different numbers of hPSCs to yield contamination frequencies of 0.03 and 0.3%. CM populations with no additional hPSCs (0%) were used as controls. The three samples were injected into the testis of male NOD/SCID/Gamma (NSG) mice (Fig. 4A) to measure teratoma potential. A small portion (50,000 cells) of each sample was used for hPSC detection by SCQC and FCM analyses (Fig. 4B). As shown in Fig. 3B, SCQC was the only technology that correctly identified the percentage of hPSCs in differentiated CMs before injection. FCM analysis failed to distinguish the differences among three samples [P > 0.05 when performing the analysis of variance (ANOVA) between any of two samples].

Fig. 4 Rare hPSCs form teratomas in vivo. (A) Workflow of the teratoma-forming assay. Exogenous rare hPSCs were spiked into hPSC-derived CMs to form cell mixtures for testicular injection. After 10 weeks, the mice were euthanized to examine teratoma formation. (B) Quantification of hPSC concentration in the samples used for injection (n = 3 for SCQC and n = 5 for FCM). (C) Representative pictures of fixed teratoma from 0% hPSCs, 0.03% hPSCs, and 0.3% hPSCs and to hPSC-derived CMs. (D) Percentage of teratoma formation in mouse models. (E) Weight of teratoma in mouse models. (F) The 0.03% and 0.30% hPSCs added to hPSC-derived CMs can form a mature teratoma that contains three germ layers, as visualized by histology. Error bar indicates the SD of the mean from all experiments (B). Whisker, box, cross, and horizontal line indicate the minimum/maximum, first/third quartile, mean, and median from each group, respectively (E). Dots represent data points (E). Cell capture experiments (B) were performed at the flow rate of 10 ml/hour using a total volume of 1 ml. Each cell suspension contained 50,000 hPSC-derived CMs spiked with various amounts of undifferentiated hPSCs in the desired final concentration, as indicated on the legend.

All the mice in both experimental groups developed teratomas after 10 weeks (Fig. 4, C and D). The averaged testis weight in the 0.03 and 0.3% hPSC group underwent a marked increase from 0.1 g to over 2 g (Fig. 4E). Conversely, mice in the control (0%) group were teratoma free, and no significant change in testis was found. This result matched with the previous studies that showed that populations consisting of 0.025% hPSCs diluted in feeder fibroblasts could initiate teratoma formation within 12 weeks (28).

We further characterized the teratomas by histopathology (Fig. 4F) and detected multiple cell types including pancreatic, respiratory, and intestinal epithelium (endoderm); cartilage, bone, fibrous, and adipose connective tissue (mesoderm); and melanocytes and glial cells (ectoderm). The histopathological finding indicates that the hPSCs retain strong pluripotency and are capable of developing mature tumors in vivo. Together, these findings demonstrate that quantification of rare hPSCs with an LOD of 0.03% or lower is required to avoid the formation of teratoma in animal models. However, the minimal number of hPSCs that is sufficient to form teratoma remains unclear and will depend on several variables including the hPSC cell line, the number of injected cells, the format of injected hPSCs (clumps or single cell), and the site of injection (29, 30). While determining this number was beyond the scope of this study, we envision being able to take advantage of the sensitivity of SCQC to determine the number of hPSCs required to form teratoma under a clinically relevant dosage.

Isolation and characterization of live rare hPSCs
Isolating live rare hPSCs in hPSC-derived CMs may provide insights into the origins of heterogeneity for in vitro differentiation processes. We next applied SCQC to the isolation and characterization of live rare hPSCs in CM populations. For these studies, we generated CM populations from HES2 and HES3-NKX-2.5GFP hPSCs using both monolayer- and embryoid body (EB)–based protocols and profiled the samples using SCQC. After capture, the external magnetic field was removed, and the cells in the chips were isolated and expanded in culture (Fig. 5A).

Fig. 5 Isolation and characterization of live rare hPSCs from manufactured batches of CMs using SCQC.
(A) Workflow of the live cell isolation. Batches of hPSCs-derived CMs were profiled using SCQC. Captured rare TRA-1-60+ cells were recovered and cultured up to 15 days for analysis. (B) Representative microscope images of colony-forming rare TRA-1-60+ cells from hPSC-derived CMs cultured in a monolayer (day 8) and as EBs (day10). The colony-forming cells maintained a high level expression of Oct4 and Nanog (n = 3 to 8). (C and D) Assessment of the pluripotency of rare hPSCs. Rare hPSCs were successfully differentiated into endoderm [FOXA2+ and SOX17+], mesoderm [SMA+ or CD144+ cells], and ectoderm [PAX6+ and Nestin+] as quantified by (C) IF and (D) FCM. (E and F) Analysis of the pluripotency-related gene expression of rare hPSCs (normal hPSCs as control). (E) Microarrayed mRNA profile of rare hPSCs (n = 4). (F) Global analysis of the state of rare hPSCs. Rare hPSCs hold a higher expression of pluripotency-related mRNA (*P < 0.05). Error bar indicates the SD of the mean from four experiments (E). Whisker, box, cross, and horizontal line indicate the minimum/maximum, first/third quartile, mean, and median from each category of genes, respectively (F).

We initiated live cell experiments by optimizing the capture and culture conditions using CM populations containing spiked pluripotent HES2 hPSCs at the frequencies of 0.01 and 0.03%. We slowed the flow rate to 4 ml/hour to secure a capture efficiency of 90 to 95%. After capture, cells were released from the SCQC chips and cultured in StemFlex medium. This medium contains bovine serum albumin (BSA) and heat-stable fibroblast growth factor (FGF), which better support the survival of rare hPSCs. After 15 days of culture, no hPSCs were detected in the negative groups that contain the cells released from the chips (fig. S7A). In contrast, we observed the formation of multiple colonies in positive groups from 0.01 and 0.03% samples (fig. S7A). It typically took 6 to 10 days to allow the rare hPSCs to recover and grow from a single cell to a colony. No noticeable internalization of TRA-1-60 MNPs was observed, and 98.5% of MNPs on the cell membrane detached in 2 days (fig. S7B). The floating MNPs were removed during regular medium change. This allows the rare hPSCs to grow in an MNP-free environment to avoid unwanted cell-MNP interaction that could hamper cell function over long time periods (31). In addition, the CMs in the negative groups were found to adhere and form a network of cells within 3 days and formed beating monolayers at day 6. This demonstrated that SCQC is a gentle cell sorting method that poses minimal stress on profiled cells.

We next proceeded to profile the differentiated batches of cardiac cells generated from monolayer-based [D4 cardiac progenitor cells (CPCs) and D8, D12, and D16 contracting CMs] and EB-based differentiation protocols (D3 CPCs and D10 and D20 contracting CMs). The phenotypes of the manufactured batches of cells were characterized as shown in fig. S7C (percentage of cTNT+ cells) and fig. S7D (representative images and video clips showing the contractility of the samples). We captured rare hPSCs in all CPC samples (three of three) (fig. S7E), most of the D8 samples (two of three) (Fig. 5B), and some of the D10 samples (two of eight) (Fig. 5B). We did not find any rare hPSCs in the D12, D16, and D20 samples. These results indicate that the rare hPSCs are mostly present in early CPCs and CMs undergoing the maturation process (D8 to D20). They also show that expression of mature cardiac markers is not an indication of a lack of rare hPSCs, as undifferentiated cells were detected in day 10 EB populations that contained greater than 75% cTNT+ cells.

Next, we characterized the pluripotency of the rare hPSCs isolated from D8 HES2 hPSC–derived CMs. At the phenotypic level, a trilineage differentiation was performed to verify the pluripotency. The rare hPSCs retained the capacity to differentiate into FOXA2+ SOX17+ definitive endoderm, SMA+ smooth muscle cells or CD144+ endothelial mesoderm-derived cells, and PAX6+ Nestin+ neural stem cells, as verified by immunofluorescence (Fig. 5C and fig. S7F) and FCM (Fig. 5D). To characterize gene expression, a quantitative polymerase chain reaction (qPCR) microarray was used to analyze the expression of key pluripotent, naïve, primed, and differentiated genes. Compared with the standard unsorted HES2, the rare hPSCs had little alteration in the expression of key genes as all fold changes remained in the range of 0.3 to 9. The highest up-regulated and down-regulated genes were EGLN1 (8.6-fold) and KHDC1L (0.37-fold), respectively (Fig. 5E). However, global analysis revealed that the rare hPSCs had statistically higher expression of pluripotent markers (P = 0.02). Together, the characterization here demonstrates the feasibility of using SCQC for identifying and isolating rare cells in hPSC-derived differentiated populations.

DISCUSSION

The SCQC method described here provides an ultrasensitive, rapid, inexpensive, and scalable means of quantifying and isolating rare hPSCs in hPSC-derived CM populations. This approach is more sensitive and cost-effective than conventional methods including FCM and ddPCR. In a manufacturing environment, SCQC provides an effective way to monitor the quality of the manufactured population with respect to the presence of contaminating hPSCs.

In addition, we found that the rare hPSCs can be detected in populations of CPCs and immature CMs using SCQC. This highlights and validates the safety concerns surrounding stem cell–based cell therapy, especially for the therapies involving progenitors and differentiated cells at the early stage. As these cells have been used in small-scale clinical trials (32, 33), the quality assessment enabled by the SCQC is critical to fulfilling the demand of safeguarding cardiovascular cell therapies (34, 35).

In general, the concept underlying SCQC is broadly applicable to all surface markers and even intracellular mRNAs via the sequence-specific MNPs clustering (36). Hence, the implementation of the SCQC can be easily extended to the quantification of other rare cells in therapeutic products or patient samples, such as circulating tumor cells and chimeric antigen receptor therapy (CAR-T). Recent work has highlighted the importance to improve manufacturing technologies to quantify rare misprogrammed leukemic B cell for safeguarding CAR-T therapy (37).

MATERIALS AND METHODS

Device design and simulation
The SCQC device implements a fluidic channel with increasing heights to generate a flow velocity gradient with eight discretized flow velocities, which correspond to eight capture zones (Fig. 1B). The height of the first zone is 50 μm, and the stepwise increment is 50 μm per zone. X-shaped structures within the microfluidic device generate capture pockets that significantly improve trapping efficiency (20). Numerical simulations of the flow velocity profiles were carried out by COMSOL Multiphysics (version 5.3; COMSOL Inc., USA) using 3D creeping flow module. The key parameters were set as below: wall condition, no slip; boundary condition, pressure of 0 Pa; suppression of backflow, yes; mesh size, physics-controlled, normal; vector field shape, normal inflow velocity; and inlet velocity rate, 3.5 mm/s. The simulated flow velocity field was processed by MATLAB R2017b (MathWorks, USA) to extract the normalized linear velocity per zone. Simulated results suggest that the normalized flow velocities range from 100 (1×) to 14% (0.14×) (fig. S1, A and B). The multidepth design has two major advantages over the previously reported planar design (19, 20, 36). First, the device remains compact when adding more zones, which reduces the fabrication cost and accelerates the microscope scan. Second, manipulating heights offers easy and fine control over the flow velocity gradient.

Design of fabrication workflow
The fabrication of multidepth microfluidic devices usually involves multiple photolithography and mask-alignment processes that markedly reduce the cost-effectiveness and scalability (38). Although three-dimensional (3D) printing has shown the potential to provide a rapid solution for fabricating multidepth microfluidic devices, existing techniques could not achieve high resolution (dot feature sizes <200 μm) in a cost-effective and robust manner (39–42). To overcome these challenges, we carefully optimized the printing conditions of a desktop stereolithographic 3D printer with a pixel size of 30 by 30 μm (fig. S1C). The 3D printer supports the formation of positive structures (i.e., microposts) compared with negative structures (i.e., microwells). The minimal printable dot and line feature is 100 and 30 μm, respectively. This optimized condition allows the successful fabrication of various positive multidepth structures with a maximal aspect ratio up to 5 (fig. S1D) within an hour at the material cost of $50. To further improve the throughput and reduce the cost, multiple molding processes have been introduced (fig. S1E). Negative molds are first generated by casting polydimethylsiloxane (PDMS) on 3D-printed positive molds. The negative molds are subsequently treated by detergent and used as a new mold to generate the microfluidic devices. In this way, one 3D-printed mold can create multiple PDMS molds for mass production. We have achieved a throughput of 40 devices per day per operator at the laboratory scale and reduced the cost to $4 per chip. The details of the X-shaped structures with high aspect ratios can be transferred properly, granting the high quality of fabricated chips (fig. S1F). The measured thickness of each zone is within ±4% of the designed thickness (fig. S1G).

Device fabrication
Positive molds were fabricated by a stereolithographic 3D printer (μMicrofluidics Edition 3D Printer, Creative CADworks, Canada) using the “CCW master mold for PDMS” resin (Resinworks 3D, Canada). The layer thickness is set to 50 μm. Negative molds were fabricated by casting PDMS (Dow Chemical, USA) on positive molds and baked at 70°C for 2 hours. Negative molds were then treated by saturated detergent solution (Sparkleen, Thermo Fisher Scientific, USA) in 70% ethanol at room temperature (RT) for at least an hour. PDMS-positive replicas were generated by casting PDMS on negative molds and baked at 70°C for 2 hours. The cured replicas were then peeled off, punched, and plasma bonded to thickness no. 1 glass coverslips (Ted Pella, USA). The bonded chips were left in a 100°C oven for 30 min to secure a robust bonding. Afterward, the silicon tubing was attached to the inlet and outlet of the device. Before use, the devices were conditioned with 1% Pluronic F68 (Sigma-Aldrich, USA) in phosphate-buffered saline (PBS) for at least 1 hour to reduce the nonspecific adsorption. Each device was sandwiched between two arrays of N52 NdFeB magnets (K&J Magnetics, USA; 1.5 mm by 8 mm) with alternating polarity. A syringe pump (Chemyx, USA) was used for the duration of the cell capture process.

Device characterization
For the characterization of microstructures, printed positive molds, PDMS negative molds, and PDMS positive replicas were sputter coated with 20-nm Au (Denton Desk II, Leica, Germany) and observed under field emission scanning electron microscopes (Hitachi SU-5000 or FEI Quanta FEG 250) using 5-kV accelerating voltage. PDMS-positive replicas were also measured by a thickness gage (Mitutoyo, Japan) to determine the thickness of each zone.

Culture of hPSC lines
 HES2 (karyotype: 46, XX) was purchased from WiCell (USA). The HES3-NKX-2.5GFP reporter cell line (karyotype: 46, XX) was provided by E. Stanley and A. Elefanty (Monash University, Australia). BYS-0113 (karyotype: 46, XY) was purchased from the American Type Culture Collection (USA). hPSCs were maintained on Matrigel (Corning, USA)– or vitronectin (Thermo Fisher Scientific)–coated well plates in feeder-free hPSC culture medium consisting of DMEM/F12 (Cellgro, Corning) supplemented with 1% penicillin/streptomycin (Thermo Fisher Scientific), 2 mM l-glutamine (Thermo Fisher Scientific), 1× nonessential amino acids (Thermo Fisher Scientific), 55 μM β-mercaptoethanol (Thermo Fisher Scientific), 20% KnockOut serum (Thermo Fisher Scientific), and rhbFGF (50 ng/ml ) (Thermo Fisher Scientific).

CM differentiation of hPSC lines
Both HES2 and HES3-NKX-2.5GFP cell lines were differentiated into CMs using a modified version of previously published cardiac differentiation protocols (21, 22). Briefly, hPSCs were grown to 80 to 90% confluence and dissociated into single cells and reaggregated to form EBs in StemPro-34 medium (Thermo Fisher Scientific) containing 1% penicillin/streptomycin (Thermo Fisher Scientific), 2 mM l-glutamine (Thermo Fisher Scientific), transferrin (150 mg/ml; Roche, Switzerland), ascorbic acid (50 mg/ml; Sigma-Aldrich), and monothioglycerol (50 mg/ml; Sigma-Aldrich), 10 mM Y-27632 (ROCK inhibitor, Tocris, UK), and rhBMP4 (1 ng/ml; R&D Systems, USA) for 18 hours on an orbital shaker. At day 1, the EBs were transferred to mesoderm induction media consisting of StemPro-34 medium with above supplements (-Y-27632) and rhBMP4, rhActivinA (R&D Systems), and rhbFGF (R&D Systems) at the optimal cardiac differentiations for each line given in fig. S6. At day 3, the EBs were harvested, washed with Iscove's modified Dulbecco's medium, and transferred to cardiac mesoderm specification medium consisting of StemPro-34 medium, 2 mM IWP2 (Wnt inhibitor, Tocris), and rhVEGF (10 ng/ml; R&D Systems). At day 6, the EBs were transferred to StemPro-34 with rhVEGF (5 ng/ml) for an additional 7 days under hypoxic conditions (5% O2). The cultures were further matured for another 8 to 10 days in StemPro-34 medium without additional cytokines under ambient oxygen conditions. At day 20, the hPSC-derived CMs were analyzed on the basis of the expression of cTNT via FCM. The EBs were cultured in ultralow attachment six-well dishes (Corning) throughout the differentiation, which routinely generated cultures with greater than 85% CMs, as determined by cTNT expression.

Definitive endoderm differentiation of hPSC lines
HES3-NKX-2.5GFP cell lines were differentiated into definitive endodermal cells using a commercially available kit (PSC Definitive Endoderm Induction Kit, A306260, Thermo Fisher Scientific). Briefly, hPSCs were seeded and grown to 10 to 20% confluence. At day 0, the medium was changed to PSC definitive endoderm induction medium A for 24 hours; after which, the medium was changed to PSC definitive endoderm induction medium B for 24 hours. The cells were then recovered for analysis and SCQC capture experiments. The differentiation routinely generates greater than 95% definitive endodermal cells based on the FCM analysis of SOX17 expression.

Generation of samples containing diluted or spiked hPSCs
Confluent hPSCs (50 to 70%) were dissociated by TrypLE (Thermo Fisher Scientific) for 3 min at RT. Dissociated cells were centrifuged, and the cell number was quantified by an automated cell counter (Countess II, Thermo Fisher Scientific) by taking the average of three to five individual counts. Low concentration solutions were achieved by serial dilution (maximal 9:1 ratio per dilution). Day 20 hPSC-derived EBs were dissociated to single cells by collagenase type 2 (300 U/mg; Worthington Biochemical Corp., USA) in Hanks’ buffer (Thermo Fisher Scientific) at 37°C for 90 min, followed by 3 min TrypLE treatment. Confluent hPSCs (50 to 70%) were dissociated by TrypLE for 3 min at RT, quantified by the cell counter, and serially diluted to achieve low concentrations of hPSCs. Populations of hPSCs and hPSC-derived CMs were combined together in the end to generate spiked samples containing 0.0005 to 5% HES2 cells in CMs. Total number of cells for each experiment is indicated in the figure captions.

Flow cytometry
For surface marker analyses, diluted or spiked samples were fixed by 4% methanol-free paraformaldehyde (PFA; Thermo Fisher Scientific) at RT for 10 min, blocked by 1% BSA (Sigma-Aldrich) in PBS (Wisent Bioproducts, Canada) on ice for 30 min, and stained by antibodies of SSEA-1, SSEA-4, TRA-1-60, TRA-1-81, CD324 (E-cadherin), CD326 (EpCAM), CD9, or CD90 (all from Miltenyi Biotec, Germany) for 10 min at 4°C in a flow buffer containing 1% BSA in PBS. For intracellular marker analyses, samples were fixed by 4% PFA at RT for 10 min, permeabilized by 0.5% Triton X-100 (Sigma-Aldrich) in PBS at RT for 10 min, blocked by 1% BSA in PBS on ice for 30 min, and stained by antibodies of SOX2, Oct3/4, Nanog (all from Miltenyi Biotec), or cTNT (BD Biosciences) for 30 min at RT in flow buffer. Detailed information regarding conjugations and dilutions is given in table S3. Stained samples were analyzed using the FACSCanto flow cytometer (BD Biosciences, USA) or the fluorescence-activated cell sorting (FACS) LSR Fortessa flow cytometer (BD Biosciences). Data were analyzed using FlowJo software (FlowJo LLC., USA). To characterize the LOD of FCM, three individual tubes were prepared for each concentration. The LOD was defined as means + 3 SD.

Droplet digital PCR
Total RNA was isolated from the spiked samples by using a single-cell RNA purification kit (51800, Norgen Biotek Corp., Canada) and used for ddPCR. The isolated RNA was used for cDNA synthesis using the First-Strand DNA Synthesis Kit (Invitrogen, USA), which contained random hexamer primers and Superscript III Reverse Transcriptase, according to the manufacturer’s protocol. The cDNA was submitted to the Centre for Applied Genomics (The Hospital for Sick Children, Toronto, Canada) for a standard ddPCR performed by a QX200 ddPCR system (Bio-Rad, USA). The TaqMan primers for target genes were purchased from Thermo Fisher Scientific: POU5F1 (OCT3/4, Hs00999634_gH), SOX2 (Hs04234836_s1), and CD326 (EpCAM, Hs00901885_m1). The TBP or B2M gene was used as the housekeeping control. The results were analyzed by the Centre for Applied Genomics using QuantaSoft Analysis Pro Software (Bio-Rad).

Characterization of magnetic labeling
Diluted samples were fixed by 4% PFA at RT for 10 min and labeled by anti–TRA-1-60 (dilution: 1:50; Miltenyi Biotec) in 1 ml of 1% BSA for 30 min at RT. Labeled samples were washed with 1% BSA in PBS twice and centrifuged at 2000 rpm for 4 min to form pellets. Pellets were then dehydrated with increasing concentrations of ethanol at 10-min intervals and embedded with Quetol-Spurr resin (Sigma-Aldrich) overnight. Samples were sliced to 70- to 80-nm-thick layers by an ultramicrotome (Ultracut RMC MT6000, Leica Microsystems, Germany) and deposited on electron microscopy grids (Ted Pella Inc.). Samples were observed under a transmission electron microscope (FEI Tecnai 20, Thermo Fisher Scientific) using 100-kV accelerating voltage.

Stem cell quantitative cytometry
Diluted or spiked samples were fixed by 4% PFA at RT for 10 min and labeled by anti–TRA-1-60 or anti-CD326 microbeads (dilution: 1:50; Miltenyi Biotec) in 1 ml of flow buffer for 30 min at RT. Labeled samples were loaded into the chips and profiled at flow rates ranging from 2 to 10 ml/hour. For the quantification of capture and depletion efficiency, captured cells were stained by DAPI and NucDead 488 (Thermo Fisher Scientific) for 10 min at the flow rate of 1 ml/hour. For the quantification of spiked hPSCs in hPSC-derived CMs, captured cells were permeabilized by 0.5% Triton X-100 in PBS at RT for 10 min at the flow rate of 1 ml/hour and stained by cocktails of antibodies [DAPI, NucDead 488, Oct3/4–PE (phycoerythrin), and Nanog–APC (allophycocyamin)] for 30 min at RT in a buffer containing 1% BSA and 0.1% Tween 20 (Bio-Rad, USA) at the flow rate of 400 μl/hour. Detailed information regarding antibody dilutions is given in table S3. After staining, the cells were washed with flow buffer for 10 min at the flow rate of 1 ml/hour. Washed chips were stored at 4°C and scanned within a week of profiling. To quantify the number of captured hPSCs, the chips were tile scanned using a Nikon Ti-E microscope with automated stages. The exposure time is 20 ms for DAPI, 10 ms for NucDead 488, 200 ms for Oct3/4-PE, and 400 ms for Nanog-APC. Scanned images were combined into a large image using Nikon NIS-Elements software (high content analysis version) and quantified using IMARIS software (Bitplane, Oxford Instrument, UK) via colocalization analysis. Cells (hPSCs) were defined as DAPI+, NucDead+, Oct3/4+, and Nanog+. To characterize the LOD of SCQC, three to five individual runs were performed for each concentration. The captured cell numbers were divided by 0.85 to normalize the effect of capture efficiency. The LOD was defined as means + 3 std.

Magnetic-activated cell sorting
For live cell separation, 1 million of HES2 hPSCs or derived CMs were labeled by anti–TRA-1-60 microbeads (dilution: 1:50) in 1 ml of flow buffer for 10 min at RT, as instructed by the manufacturer. Labeled samples were applied to MS columns (Miltenyi Biotec) and washed twice using the flow buffer. TRA-1-60–positive cells were recovered from the column by firmly pushing the plunger into the column twice. Recovered cells were centrifuged and immediately processed for cell counting using an automated cell counter. For stained cell separations, 1 million of HES2 PSCs or derived CMs were fixed, permeabilized, and stained by DAPI, Oct3/4-PE, and Nanog-APC. These cells were subsequently labeled with anti–TRA-1-60 microbeads (dilution: 1:50) in 1 ml of flow buffer for 10 min at RT. Labeled samples were then sorted using MS columns. Recovered cells were centrifuged and immediately processed for cell counting.

Teratoma formation and analysis
All animal experiments were carried out in accordance with the protocol approved by the University of Toronto Animal Care Committee. Male NOD/SCID/interleukin 2 receptor Gamma chain null (NSG) strains of mice at 6 to 8 weeks of age were purchased from the Jackson laboratory (USA) and maintained at the University of Toronto animal facility. Spiked sample with 1 × 106 cells in 15 μl of Matrigel (Corning) was injected into the pericardium of testis. Ten weeks after injection, mice were euthanized, and the formation of teratomas was examined. Extracted teratomas were weighed and fixed in 10% formalin (Sigma-Aldrich). Formalin-fixed, paraffin-embedded teratomas were sectioned (5-μm thickness) and stained with hematoxylin and eosin. Histological examination was performed by a licensed veterinary pathologist blinded to the difference of samples to identify the germ layers in the teratomas.

In vitro colony-forming assay for spiked samples
Spiked samples were labeled by anti–TRA-1-60 microbeads in 1 ml of 1% BSA in PBS for 30 min at RT. Labeled samples were loaded into the chips and profiled at flow rates of 4 ml/hour. After profiling, magnets were removed from the chip. The negative groups were obtained from the syringe, and the positive groups were obtained by withdrawing cells in the chips with a new syringe. The profiled groups were centrifuged and resuspended in the hPSC culture medium for reculturing on vitronectin-coated well plates. At days 3, 6, 10, and 15 after profiling, the recultured cells were fixed by 4% PFA at RT for 10 min, permeabilized by 0.5% Triton X-100 in PBS at RT for 10 min, and stained by antibodies of DAPI, TRA-1-60-Vio488 (Miltenyi Biotec), Oct3/4-PE, and Nanog-APC for 60 min at RT in flow buffer. Detailed information regarding conjugations and dilutions is given in table S3. The plates were tile scanned using the Nikon Ti-E microscope. The exposure time is 20 ms for DAPI, 100 ms for TRA-1-60-Vio488, 200 ms for Oct3/4-PE, and 400 ms for Nanog-APC. hPSCs were defined as DAPI+, NucDead+, Oct3/4+, and Nanog+. Number of colonies (>4 hPSCs per colony) per well was quantified manually.

Isolation and characterization of rare hPSCs isolated from manufactured batches
In addition to the abovementioned EB-based protocol, a commercially available cardiac differentiation kit (A2921201, Thermo Fisher Scientific) was also used to generate batches of CMs from a monolayer. Briefly, HES2 and HES3-NKX-2.5GFP cells were maintained in vitronectin-coated well plates in Essential 8 medium (Thermo Fisher Scientific) for 2 days (days −2 to 0). The confluency of cells at day 0 is between 50 and 70%, as suggested by the manufacturer. Then, the medium was replaced with CM differentiation medium A and B at days 0 and 2, respectively. At day 4, the medium was changed to Cardiomyocyte Maintenance Medium (A2920801, Thermo Fisher Scientific) and changed every 2 days until day 16. Contracting CMs appeared at day 8, and the beating conditions of the CMs were monitored at days 8, 12, and 16 using the Nikon Ti-E microscope. The percentages of cTNT-positive cells at days 8, 12, and 16 were quantified by FCM using the protocol described in the “Flow cytometry” section. At days 4, 8, 12, and 16, monolayers of hPSC-derived CMs were dissociated by TrypLE for 4 min. At days 10 and 20, hPSC-derived CMs grown as EBs were dissociated to single cells by collagenase type 2 (300 U/ml) in Hanks’ buffer at 37°C (30 min for day 10 and 90 min for day 20), followed by 3-min TrypLE treatment. Dissociated cells were labeled by anti–TRA-1-60 microbeads and profiled using the protocol same to the spiked samples. The positive groups were recultured in StemFlexTM medium (A3349401, Thermo Fisher Scientific). At days 2, 6, and 10 after profiling, the positive cells were fixed, stained, and quantified using the same protocol used for the spiked samples. If rare hPSCs (DAPI+, TRA-1-60+, Oct3/4+, and Nanog+) were found at day 2, the same groups were passaged when it reached 80 to 90% confluency up to 14 days to allow rare hPSCs to proliferate. To examine the pluripotency of isolated rare hPSCs, proliferated hPSCs were differentiated into three germ lineages using a trilineage differentiation kit (130-115-660, Miltenyi Biotec), which typically takes 7 days. At day 7, the cells were fixed, permeabilized, and stained with DAPI, FOXA2, and SOX17 (for endoderm); DAPI, smooth muscle actin (SMA), and CD144 (for mesoderm); and DAPI, PAX6, and Nestin (for ectoderm). Detailed information regarding conjugations and dilutions is given in table S3. The stained plates were observed using the Nikon Ti-E microscope. The exposure time is 20 ms for DAPI, 100 ms for FOXA2-PE, 400 ms for SOX17-AF647, 20 ms for SMA-PE, 400 ms for CD144-AF647, 200 ms for PAX6-PE, and 600 ms for Nestin-AF647. To examine the naïveness of isolated rare hPSCs, total RNA was isolated from the proliferated hPSCs following the same protocol used for ddPCR. A comparative CT experiment was performed on an Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific) using hPSC naïve-state qPCR array (07521, Stemcell Technologies, Canada). The assay was carried out using 5 μl of TaqMan Universal Mix, 4 μl of nuclease-free water, 1 μl of cDNA (10 ng/μl) for each sample in a 96-well plate. Cycling conditions for the qPCR were 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The post-PCR analysis was performed by an online tool provided by the manufacturer (https://stemcell.shinyapps.io/qpcr_tool/).

Acknowledgments

We would like to thank members of the Kelley and Keller laboratory, especially B. Green, X. Fan, A. Garcia, S. Ogawa, and S. Protze, for experimental advice and critical comments on the manuscript, A. Elefanty and E. Stanley (Monash University) for providing the HES3-NKX-2.5GFP reporter cell line, M. Ly, K. Patel, and H. Patel (Creative CADworks) for establishing the protocol for 3D printing, T. Paton (The Hospital for Sick Children) for assistance in ddPCR, M. Ganguly (University Health Network) for assistance in histology, and M. Larsen (Mbed Pathology) for assistance in pathology. Funding: Research reported in this publication was supported in part by the Canadian Institutes of Health Research (grant FDN-148415 to S.O.K. and grant FDN-159937 to G.M.K.). This research is part of the University of Toronto’s Medicine by Design initiative, which receives funding from the Canada First Research Excellence Fund. Z.W. was supported by a Connaught International Scholarship. Author contributions: Z.W., M.G., E.H.S., G.M.K., and S.O.K. conceived and designed the experiments. Z.W., M.G., R.M.M., S.U.A., M.L., L.Z., S.P., and Y.Z. performed the experiments and analyzed the data. All authors discussed the results and contributed to the preparation and editing of the manuscript. Competing interests: G.M.K is a founding investigator, equity holder, and a paid consultant for BlueRock Therapeutics LP and a paid consultant for VistaGen Therapeutics. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from S.O.K.

A rapidly prototyped lung-on-a-chip model using 3D-printed molds

by

A rapidly prototyped lung-on-a-chip model using 3D-printed molds

Jesus Shrestha, Maliheh Ghadiri, Melane Shanmugavel, Sajad Razavi Bazaz, Steven Vasilescu, Lin Ding and Majid Ebrahimi Warkiani

Organ-on-a-chip is a microfluidic cell culture model that replicates key organ-specific microarchitecture and pathophysiology in vitro. The current methods to fabricate these devices rely on softlithography, which is usually tedious, laborious, and requires adroit users as well as cleanroom facilities. Recently, the use of 3D-printing technologies for the rapid fabrication of molds for polydimethylsiloxane (PDMS) casting is on the rise. However, most of the 3D-printed materials are unsuitable for PDMS casting. To address this issue, we have improved the existing techniques and introduced a modified protocol for the surface treatment of 3D-printed molds, making them ideal for repeated long-term PDMS casting. Using this protocol, we have fabricated a simple open well lungon-a-chip model to simulate the in vivo environment of airway at air-liquid interface under dynamic condition. To validate the functionality of the developed chip, Calu-3 cells were cultured in the chip and maintained at an airliquid interface. The model demonstrated that the cultured cells replicated the 3D culture-specific-morphology, maintained excellent barrier integrity, secreted mucus, and expressed cell surface functional P-glycoprotein; all indicative of a promising in vitro model for permeability assays, toxicological tests, and pulmonary drug delivery studies. To validate the suitability of this lung-on-a-chip in vitro model, the effects of cigarette smoke extract (CSE) on Interleukin-6 (IL-6) and Interleukin-8 (IL-8) release from cultured Calu-3 cells were examined. CSE treated cells showed significantly higher secretion of IL-6 and IL-8 over 24 h compared to the cells treated with both CSE and Budesonide, an anti-inflammatory drug. Moreover, our results illustrated that CSE reduced the expression of Ecadherin as an adherent junctional protein. In conclusion, the proposed protocol demonstrated an easy and lowcost fabrication technique which will allow a biologist with minimal technical skills to rapidly prototype molds for different/versatile organ-on-a-chip models

We kindly thank the researchers at University of Technology Sydney for this collaboration, and for sharing the results obtained with their system.

Introduction

The burden of respiratory diseases throughout the world is on the rise, with almost 4 million premature deaths from chronic respiratory diseases every year, indicating a serious health issue (Forum of International Respiratory Societies, 2017). Therefore, there is an urgent need to develop new respiratory drugs, which require developing better and physiologically relevant respiratory disease and drug testing models (Shrestha et al., 2020). The conventional methods used to reproduce the microenvironment, functions, and physiology of the human lung are two dimensional (2D) monolayer cell culture and transwell-based immersion cultures (Huh et al., 2011). However, these models do not accurately replicate in vivo three-dimensional (3D) cellular structure, cell-cell interactions, the air-exposed environment of the alveolar cells, and the physiological functions at the organ level.

The inability of existing cell culture models to accurately and reliably model respiratory disease has led to the emergence of microfluidic cell culture systems. Microfluidic systems precisely replicate the physiological conditions required for both basic research and drug development, enabling drug discoveries through systematic testing (Neuzil et al., 2012; Azadi et al., 2019). Several microfluidic organ-on-a-chip models replicating the 3D microarchitecture and mechanical and physiological features of different organ tissues have been developed (Esch et al., 2015). However, conventional softlithography techniques used to fabricate organ-on-a-chip devices impose severe limitations on time, geometric complexity, and cost, often requiring a separate cleanroom facilities and adroit users, as a result of which, impedes the pace of development and innovation in microfluidic applications. Therefore, a method for rapid fabricating microfluidic devices with little technical expertise would be ideal for scientists with little knowledge of microfabrication.

Recently, additive manufacturing has emerged as an alternative for the fabrication of microfluidic devices (Vaezi et al., 2013; Erkal et al., 2014; Bhattacharjee et al., 2016). With further developments in 3D-printing, there has been a significant increase in the utilization of 3D-printed part as molds to generate PDMS replicas. 3D-printing has superiority over other techniques of mold fabrication such as micro-milling or laser cutting, enabling higher resolution features and a control over height parameters (Guckenberger et al., 2015; Condina et al., 2019). Among all 3D printing methods, stereolithography apparatus (SLA) and digital light processing (DLP) offer great advantages and are therefore preferred for microfluidics and biomedical applications (Macdonald et al., 2017). This allows researchers to rapidly design and alter complex microstructures without expending large amounts of time or resources. In addition, 3D-printed molds are more adaptable as they can include multiple design features at different channel heights not possible using conventional photolithography techniques. However, most of the 3D-printed molds printed via SLA/DLP techniques are not immediately suitable for PDMS casting since residual monomers and oligomers on the surface of the 3D-printed parts impede PDMS polymerization. Several groups have proposed surface treatment methods for 3D-printed molds to make it suitable for casting PDMS, including ink or lubricant infused coatings, plasma treatment, salinization, and heat cycling (Waheed et al., 2017; Chan et al., 2015; Comina et al., 2014; Villegas et al., 2018). Reported protocols are time-consuming, labor-intensive, and lack reproducibility (Razavi Bazaz et al., 2019). Although many protocols contain similar basic parameters such as treatment time, curing temperature, and UV exposure, these parameters are subject to change according to feature dimensions. Some groups use the services provided by commercial 3D-printing companies to manufacture their molds (Park et al., 2019; Novak et al., 2018). However, these molds can be expensive and have waiting periods associated with manufacturing and delivery (Ellison et al., 2016). This type of arrangement does not allow rapid prototyping to be performed, where quick changes in design parameters are required. Curing temperature of PDMS on the 3D-printed molds is a vital step that requires careful optimization to prevent material strain and microstructure deformation. Therefore, it is crucial to develop an optimized surface treatment process for 3D-printed molds ensuring long-term cell viability in the organ-on-a-chip devices.

Herein, we present an enhanced protocol for surface treatment of 3Dprinted molds to fabricate a simple open access lung-on-a-chip design. The protocol presented here enables the quick fabrication of molds for long-term use without the development of any cracks or channel deterioration through carefully optimized steps and parameters. This treatment process allows high-resolution repetitive PDMS casting using 3Dprinted molds. The fabricated chip was further optimized by testing different membranes and ECM coatings for cell growth and extended viability. The chip allowed the lung epithelial cells to be cultured at an air-liquid interface under dynamic conditions; the transparency of PDMS enabled real-time cell visualization and chip monitoring (Fig. 1). Calu3 cells are known to highly express the tight junction proteins Occludin and E-cadherin, which makes it a suitable cell type for analyzing tightjunction formation and cell barrier functions (Kreft et al., 2015; Haghi et al., 2010). Mucus production, differentiation, and the expression of transport proteins are other features of Calu-3 cells, making them useful for modelling the airway epithelium. Different study groups have already shown the suitability of this cell line to use it as a respiratory in vitro model (Zhu et al., 2010; Foster et al., 2000; Florea et al., 2003). Thus, the Calu-3 cell line was chosen for our lung-on-a-chip device. Using Calu-3 cells, we demonstrate the versatility of our lung-on-a-chip model through the assessment of CSE effects and Budesonide treatment on the secretion of inflammatory markers and cellular expression of the junction protein E-cadherin. Furthermore, we provide a functional analysis of the epithelium cell layer generated. The flexibility of direct 3D-printing utilized in this study will aid the fabrication of novel organ-on-a-chip designs within a short time frame. The printing process we report here is versatile enough to be adapted for multiple organ-on-a-chip models beyond what is studied here.

Fig. 1. Microfluidic model of human Lung-on-achip design and fabrication: A) A conceptual schematic of the experimental setup showing the human respiratory system B) A cross-section of human airway tissue. C) 3D printer was used to fabricate the open well design of the chip model with upper and lower layers to recapitulate human lung. D) The top layer contains a central open well for cell seeding, and an inlet and an outlet for media in the lower channel. The lower layer includes a channel for media flow. The porous PC membrane is carefully placed and aligned between the two layers, where the cells attach and grow. E) Once the cells were confluent, the effects of CSE on the cells were analyzed.


2. Materials and methods
2.1. Mold fabrication and surface treatment
The mold was fabricated using a digital light processing (DLP) 3Dprinter, (MiiCraft Ultra 50, MiiCraft, Hsinchu, Taiwan) with a printing area of 57 X 32 X 120 mm and XY resolution of 30 μm. The printer projects a 385–405 nm UV wavelength through the resin (BV-007) on the resin bath. The design process began with Computer-Aided Design (CAD) modelling of the required geometries in SolidWorks (2016), after which designs were exported in an STL file format to the Miicraft printer software (MiiCraft 125, Version 4.01, MiiCraft Inc). To enable highresolution printing of mold features, the print options were carefully tailored to each design. Smaller design features required a slice thickness of 10 μm and a curing time of 1 s per slice, where features were less fine, slice thicknesses of 30 and 50 μm were used. Considering the size of the printed molds, a base layer was used to ensure the part adhered to the picker for the duration of the print. The curing time for the base layer was set to 24 s. A buffer layer was used to facilitate the transition from the base layer into the printed part.

To prevent the PDMS from sticking to the 3D-printed mold, surface treatment of the resin mold is mandatory. We have optimized the surface treatment method proposed before (Waheed et al., 2017) to shorten the overall duration, making it suitable for channels with smaller dimensions and with the ability of repeated casting without affecting the mold structures. First, the 3D-printed mold was washed with isopropanol (IPA) followed by high-pressure air drying after the print. After that, the post-curing of the mold is required for 3 min (steps of 10 s). To make sure that uncured monomers and oligomers on the surface of the mold are eliminated, the prepared mold was then soaked inside 100% ethanol for 2 h. The surface of the mold must be prepared for the step of an easy detachment of the PDMS; hence, Oxygen plasma treatment (Basic Plasma cleaner PDC-002, Harrick Plasma) was carried out for 2 min. In the end, surface of the mold was silanized using trichloro (1H, 1H, 2H, 2H-perfluoro-octyl) silane (Sigma-Aldrich, Australia) in a desiccator under vacuum for 1.5 h. The workflow for mold design and fabrication is illustrated in Fig. 2.

2.2. Flow simulation and diffusion in the channels
To characterize how the microchannel impacts fluid flow and its effect on the membrane, a 2D model of the proposed design was analyzed using Comsol multiphysics 5.3a, a commercial CFD package (Bazaz et al., 2018). To study the effects of fluid flow on the membrane, a coupled equation in free and porous media flow is mandatory. The flow in the microchannel is described using continuity and Navier-Stokes equations.

Here, u represents the vector of fluid velocity, ρ denotes fluid density, μ refers to dynamic viscosity, and p is the pressure. In this study, Brinkman equations are used for the momentum transport over the membrane.

where εp is porosity, κ is permeability, Qbr denotes mass source or mass sink, and F equivalents to the external forces. Using an oversimplified Darcy's law results in the neglection of the viscous effect which arises from the fluid media. As an alternative, the Brinkman equation was employed to solve the fluid within the porous media. A hydrophilic Isopore polycarbonate membrane (Sigma-Aldrich, Australia) is used in this study. The pore size of the membrane (dpÞ is 0.4 μm while the porosity ðεp Þ is 15%, and membrane thickness is 10 μm. The membrane permeability can be calculated via a packed bed model, and its value

Fig. 2. Protocol for the surface treatment of the 3D printed resin molds: A) CAD design of the desired mold using Solidworks. B) The finalized designs are printed using the 3D printer. C). The mold is then washed with IPA, followed by UV curing. D) and E). The mold is then dipped in ethanol before plasma treating. F) and G) Before casting PDMS, the mold is silanized. Once the PDMS is cured, it is carefully cut out from the molds to form two slabs. The PDMS slabs are then cleaned with alternative IPA and Ethanol washing before being plasma treated, aligned, and bonded carefully to make a complete lung-on-a-chip adjusted to 4.98e-18 m2 and 4.98e-19 m2 in the absence and presence of the cells on the membrane, respectively. The physics-controlled mesh was used as the domain grid in this study. The normal inflow velocity was applied at the inlet, while zero static pressure was considered at the outlet.

2.3. Device fabrication
Following surface treatment of the resin mold, a mixture of PDMS base and curing agent (ratio of 10:1) was prepared (Sylgard 184 from Dow Corning, MI, USA) (Fig. 2). The mixture was then degassed in a desiccator for 20 min and poured into the treated molds. The PDMS was left to cure in the hot air oven for 4–5 h at 45 C. The PDMS layer was gently lifted off from the mold, trimmed to desired shapes, and holes are punched for inlets and outlets. The PDMS layers were thoroughly cleaned with IPA and ethanol, at least three times, followed by air drying between each wash. Polycarbonate membrane was cut into a square shape, large enough to cover the central circular well and carefully placed on the lower PDMS layer. The contact surface of both the upper and lower PDMS layers was plasma treated for 1 min and aligned precisely to ensure perfect bonding. The bonded chip was then kept in the oven again for 4 h at 45 C to bond, followed by testing for leakage using food dyes.

2.4. Cell culture
Human airway epithelial cell line, Calu-3 from American Type Cell Culture Collection (ATCC, Rockville, IN, USA) were cultured in Dulbecco's Modified Eagle's medium: F-12 (DMEM: F12) supplemented with 10% fetal bovine serum (FBS), 1% L-glutamine, 1% penicillin ( Gibco, Life Technologies, Australia) and 1% non-essential amino acids (SigmaAldrich, Australia). The chip was sterilized by flushing 70% ethanol through the channels, followed by hot air drying in the oven. It was then exposed to UV light in a biosafety cabinet for 30 min before the extracellular matrix (ECM) coating. Once the membrane was coated with ECM, the chip was kept in a humidified incubator at 37 C with 5% CO2 for 2 h before seeding cells. Residual ECM coating in the channels was removed by passing fresh culture media. After seeding with cells, the chips were maintained in 37 C and 5% CO2. The media in both the upper and lower channels were replaced daily until the cells were confluent (day 5). Once the confluency was achieved, the media from the upper channel was aspirated to allow cells to grow in the air-liquid interface. The lower channel was then attached to a syringe pump (Fusion 200, Chemyx Inc.) with a flow rate of 30 μl/h to maintain a dynamic condition with fresh media flow.

2.5. Cell adhesion to the membrane and ECM coating
Different combinations of membranes and ECM coatings were tested to guarantee optimal cell growth and the formation of a confluent monolayer. The different membranes tested were Isopore membrane Polycarbonate filter (PC) (Merck, Australia), Nuclepore track-etch PC membrane (Whatman, Australia), Polyester (PE) from transwell cell culture inserts (Corning Incorporated, Australia), and Millipore from Millicell culture plate inserts (Sigma- Aldrich, Australia), all with 0.4 μm pore size. The membranes were first tested without any ECM coating. Calu-3 cells were seeded at 8000 cells/mm2 and incubated at 37 C, 5% CO2. Media in both channels were changed every day. The cells were stained with a working concentration of 1 μg/ml of Hoechst 33342 (Abcam, Australia) and incubated for 15 min on day 5. After washing with phosphate-buffered saline (PBS), it was observed under the microscope to identify the most suitable membrane with the highest level of cellular attachment. To further enhance the cell attachment, selected membrane was coated with different ECMs: (i) 5 μg/ml of Fibronectin (Corning, 356008), (ii) 3 μg/μl of Collagen-I (Col-I) (Corning, 354236), (iii) 6 μg/μl of Matrigel (Corning, 354234) and (iv) a mixture of 6 μg/μl Matrigel and 3 μg/μl Col-I. This was followed by incubation for 2 h at 37 C, 5% CO2, and washing with PBS (Sigma- Aldrich, Australia) to get rid of excess ECM. Cells were seeded and incubated. The media from both channels was changed the next day. The cells were then stained with Hoechst 33342 on day 5 and observed with Olympus Ix73 Inverted Microscope for cellular attachment to determine the optimal ECM coating for Calu-3 cell culture.

2.6. Viability and functionality testing of the device
After the selection of the membrane and ECM coating, the following experiments were conducted in the final chip design.

2.6.1. Cell viability and proliferation
Live/dead cell double staining kit (Sigma-Aldrich, Australia) was used to distinguish viable cells from non-viable cells. The assay solution of the stain was prepared by adding 10 μl of Calcein-AM (Solution A) and 5 μl of Propidium Iodide (Solution B) in 5 ml of PBS. The cells were washed with PBS before adding the assay solution. 30 μl of the assay solution was added to the cell layer in the upper channel and incubated at 37 C for 15 min. Olympus Ix73 Inverted Microscope was used to simultaneously observe live and dead cells.

2.6.2. Mucus staining
The mucus production from the Calu-3 cells in the chip was characterized on day of 3, 5, 7, 9, and 11 of culture by staining the glycoprotein in the mucus using Alcian blue (1%. (w/v) in 3% (v/v) acetic acid/water at pH 2.5) (Sigma- Aldrich, Australia). The monolayer of the cells was washed twice with PBS. After fixing the cells with 4% (v/v) Paraformaldehyde for 20 min, PBS wash was repeated. Finally, 50 μl of Alcian blue stain was added. The chips were incubated for 15 min and washed multiple times with PBS until the rinsate was clear. Images were obtained using an Olympus Ix73 Inverted microscope. Depending on the ratio of red, green, and blue (RGB) from the microscopic images of mucus staining, data were analyzed to generate a semi-quantitative estimate of mucus concentration (Haghi et al., 2010). The mean RGB values were obtained with Image J (v1.52p, NIH) with Color Profile (Dimiter Prodanov; Leiden University Medical Centre, Leiden, Netherlands). The mean RGBB was divided by the total sum of RGB values for each image (RGBR þ RGBG þ RGBB) to calculate the ratio of blue (RGBB ratio). The mean RGBB of eight images was used to quantify the secretion of mucus by the Calu-3 cells for days 3, 5, 7, 9 and 11.

2.6.3. Paracellular permeability of sodium fluorescein (Flu-Na)
 The barrier integrity of the cells grown on the membrane was assessed by using Flu-Na (MW 0.367 kDa, Sigma-Aldrich, Australia) on day 7. The cell layer in the upper channel was washed with warm PBS (37 C) after removing the media from both the channels. The lower channel was filled with prewarmed PBS, and the cells were incubated at 37 C for 2 h. The upper channel was filled with 50 μl of flu-Na solution (2.5 mg/ml), while PBS was flushed through the lower channel at 0.5 μl/min for both blank devices and seeded chips. Samples were collected from the lower channel every 30 min for a total of 120 min. The fluorescence of flu-Na present in each sample was measured in Corning full black clear bottomed 96-well plates using a fluorescence plate reader (Infinite 200 PRO; TECAN), using excitation and emission wavelengths of 485 and 520 nm, respectively.

2.6.4. Flow cytometry detection of cell surface P-Glycoprotein
On day 7, the Calu-3 cells were harvested from the membrane of the chip by trypsinization (TrypLE Express; Gibco). After washing twice with PBS, the cells were labelled using 20 μl of FITC-anti P-gp (clone17F9, BD Pharmingen, USA). It was then incubated in the dark at room temperature for 30 min. The cells were washed twice again with PBS before resuspending them in 200 μl of PBS, spiked with (1 μg/ml) propidium iodide (PI) (Sigma- Aldrich, Australia). Samples were analyzed by flow cytometry (FCM) using the CytoFLEX LX (Beckman Coulter, Life Sciences, USA) and CytExpert Software. 

2.7. Preparation of CSE and treatment of the cells
Cigarette smoke extract (CSE) was prepared by a method modified from a publication (Laurent et al., 1983). One Marlboro Red cigarette (Philip Morris, Victoria, Australia) was bubbled through 25 ml of DMEM in a T-75 flask at a constant rate, which was considered as 100% concentration CSE. The collected CSE was filtered and diluted to the required CSE concentrations in the media. Prepared CSE was immediately used and diluted within 30 min. To stimulate Calu-3 cells with a non-toxic concentration of CSE, first cytotoxicity assay was performed. Briefly, Calu-3 cells were cultured in 96-well plates were stimulated with serial dilutions of CSE from 0.0675% to 100%. The cytotoxic concentration of CSE on Calu3 cells over 48 h was tested using MTS assay kit (Promega, CellTiter 96® AQueous One Solution Cell Proliferation Assay- Australia). Then, cells grown on the chip were stimulated with appropriate concentration of CSE (the concentration below IC50 (Inhibitory concentration at 50%). The experiment was categorized into four groups of chips, and the cells were grown as mentioned previously for 7 days. One group was first treated with 100 nM Budesonide for 24 h from the lower chamber and then treated with CSE from the top (Bud-CSE). The second group was treated with CSE and 100 nM Budesonide at the same time (CSE-Bud). The third group was only treated with CSE and the fourth group was the control (no treatment). The lower channel was connected to a syringe pump and media was collected in a tube for 24 h; this media was used to measure IL-6 and IL-8 with ELISA technique. The media collected from the chips were stored at -80 C.

2.8. Enzyme-linked immunosorbent assay (ELISA) for IL-6 and IL-8
The media collected from the chips were stored at 80 C until performing the experiment. After thawing them, the levels of secreted IL-6 and IL-8 were analyzed according to the manufacturer's instructions, using commercial human IL-6 and IL-8 ELISA kits (BD Pharmingen, San Diego, CA, USA). The absorbance was read at 450 nm/570 nm using plate reader (Infinite 200 PRO; TECAN).

2.9. Immunofluorescence staining
To visualize the effects of CSE and Budesonide treatment on the tight junctions, the Calu-3 cells were assessed by imaging of immunolabelled stains of tight junction protein, E-cadherin. Images were taken using a Nikon A1 confocal microscopy (Japan). All immunostaining steps were conducted at room temperature. The cells were fixed with 4% volume/ volume (v/v) paraformaldehyde (Sigma- Aldrich, Australia) in PBS after washing three times with PBS. It was then incubated for 15 min and washed with PBS twice. This was followed by permeabilization of the cell membranes with 0.1% (v/v) Triton X 100 (Sigma- Aldrich, Australia) for 10 min and blocking with 1% (w/v) bovine serum albumin (BSA) (SigmaAldrich, Australia) in PBS for 1 h. After further washing with PBS, cells incubated with 50 mM ammonium acetate in PBS for 10 min. Then cells were washed with PBS and incubated at 37 C with CD324 (E-Cadherin) monoclonal antibody (10 μg/ml in PBS) (Invitrogen) for 1 h. Rewashing with PBS was done before adding AlexaFlour 594 goat anti-mouse IgG1 (10 μg/mL in PBS) (Invitrogen, Australia) to incubate for 1 h. After washing with PBS, the cells were counterstained with 1 μg/mL 40 , 6-diamidino-2- phenylindole (DAPI) in water for 10 min before washing with PBS. The membrane was carefully cut from the open well of the chip and mounted on a microscopic glass slide and covered by a coverslip. Care was taken to prevent curling and tearing of the membrane, and the slides were stored at 4 C. The slides were viewed the next day using the Nikon A1 Confocal Laser Microscope with NIS-Elements C Software.

2.10. Statistical analysis
Data were analyzed using IBM SPSS Statistics 25 software (USA). The ANOVA one-way analysis was used to determine significance (P < 0.05). All results are expressed as the mean standard deviation (SD) of at least three independent determinants.

3. Results and discussion

3.1. Fabrication of the mold and the device
Lung-on-a-chip models typically have closed system designs with straight channels (Huh et al, 2010, 2012, 2013; Jain et al., 2018; Benam et al., 2016). Although this allows uniform flow, manipulating cell suspensions and micro-volumes of fluids through microchannel is challenging. Functional tests such as immunostaining, mucus secretion, and permeability assays become difficult to perform within a closed design chip. The open well design of our device allows easy access to the membrane for uniform coating, cell seeding, fluid manipulation, and sample collection. Moreover, the open access design allows the cells to be directly exposed to the CSE, drugs or nanoparticles, making transport or migration studies more controllable. Multiple open well chambers can be interconnected to each other to conduct independent as well as parallel studies. This was first suggested by Blume et al. who developed a similar open well design compatible with commercially available Transwells to be interconnected. (Blume et al., 2015). However, this is a complicated process compounded by the number of components comprising their chip design. We have effectively simplified the lung-on-a-chip model to facilitate the on chip cell culture with the potential for multichip interconnections. The simple maintenance and usability of our chip will allow a person familiar with conventional cell culture methods to conduct their experiments in a more relevant microenvironment, even with minimum microfluidic knowledge.

The open well lung-on-a-chip model presented here consisted of an upper PDMS layer with a large circular well, an inlet, and an outlet on either side (Fig. 3A). The lower layer comprised of a central chamber connecting to two straight channels from either side. The thin, porous PC membrane separated the two PDMS layers. The 3D-printed mold consisted of raised channels to imprint the PDMS surface. Surface treatment of the mold before casting PDMS is vital for PDMS casting (Chan et al., 2015). This treatment cures the surface areas of the mold left uncured from the 3D-printing process. To achieve this, the mold must be free of any debris, residual monomers or oligomers. The process of silanization provides a hydrophobic fluorinated monolayer on the 3D-printed mold that prevents the sticking of the PDMS to the resin mold. This methodology allows easy peeling off the PDMS from the mold. Curing PDMS at 45 C for 4–5 h was found to yield the best surface finish for the 3D-printed resin molds. PDMS incubation with the mold at higher temperatures induced the formation of surface fractures within the mold, negatively impacting the surface finish of the PDMS piece, resulting in leaking of the bonded device. Hence, it can be reasoned that the temperature has a significant influence on the molds.

This approach of fabricating and treating the 3D-printed resin molds is a simple, cost-effective, and time-efficient method for producing 3D microfluidic lung-on-a-chip models compared to the conventional softlithography techniques using silicon wafers. The 3D-printing of the molds, surface treatment, and chip fabrication can be completed within a day. This helps to avoid the typically lengthy processing and delivery time of the commercially manufactured molds. This time reduction is a result of optimized fabrication steps minimizing the risk of human error. Thus, the approach presented here enables researchers to rapidly prototype multiple designs of different geometries within a short period. The molds created can be used repeatedly without concern over the reproducibility between chips. However, despite these advantages, 3D-printing has certain limitations regarding the resolution of printable features and quality of the surface. For instance, ink-jet type 3D-printing is an option available for 3D-printing molds to cast PDMS and fabricate chips (Kamei et al., 2015). Nevetheless, ink-jet 3D-printing can result in increased surface roughness and may cause altercations in channel profiles.

Fig. 3. Simulations of fluid flow in the lung-on-a-chip model: A) Geometry and boundary conditions used for computational modelling. B) Velocity profile along the length of the microchannel. It is evident that the velocity distribution across the upper channel is small enough to not negatively impact attached cells. C) Velocity distribution along the length of the channel. It is shown that the velocity profile in lower channel has a parabolic profile. D) Velocity profile at the upper channel. The order of velocity is small enough so that cells are not influenced by shear rate.

Materials

M50

Learn More

3.2. Fluid behaviour and diffusion simulation in the microchannel
To better quantify the fluid behaviour within the lung-on-a-chip device, the flow velocity was simulated, as shown in Fig. 3B. The numerical results reveal that the flow exists at a steady-state condition, and the relative velocity is higher in the lower channel compared to the upper channel. Streamlines in the lower channel run along the length of the channel from inlet to outlet and shows that the velocity profile in the lower channel is parabolic (Fig. 3B) while, in the upper channel, fluid is driven from the lower channel across the porous membrane before exiting from the membrane. These results indicate that fluid can traverse the membrane, passing between the upper and lower channels continuously. Thus, the velocity distribution along the length of membrane is calculated and illustrated in Fig. 3D. According to the value of velocity in the upper channel, the effect of media flow on the cells growing on the membrane is negligible. This allows the secreted molecules from the cells cultured on the membrane to remain undisturbed in the interstitial fluid since the flow of media is limited to the lower surface of the cell layer (Nalayanda et al., 2009; Walker et al., 2004). This was further validated by mucus staining performed over several days.

3.3. Cell adhesion to the membrane and ECM coating
To determine the optimal membrane and ECM coating combination to support the adhesion and growth of the Calu-3 cells in the device, several types of membranes and ECMs were tested. We evaluated the results based on the cell adhesion and area coverage by cells. Chips were seeded with 8000 cells/mm2 which ensured a viable cell population covering a large area. Based on these findings, 8000 cells/mm2 seeding density was used for all subsequent experiments in this study. On the fifth day of incubation, the chip was observed for cell attachment; to help confirm cellular attachment, the cells were stained with Hoechst (Fig. 4AI). The PE membrane consistently showed large empty areas with minimal cellular attachment. In comparison the Millipore membrane provided greater cell attachment, but the membrane with the most consistent cell attachment was PC (Merck, Australia). Therefore, for further experiments on ECM coatings, PC (Merck, Australia) membranes were used.

To identify which ECM coating achieved the highest level of cellular attachment, cell viability and the formation of a monolayer, multiple PC membranes were coated with Fibronectin, Collagen-I, Matrigel, and a mixture of Matrigel and Collagen-I. A mix of 6 μg/μl Matrigel and 3 μg/μl Col-I was selected based on findings suggested by Humayun et al. (2018). Seeded membranes were then incubated and washed with PBS to remove all detached and dead cells. The cells were then stained with Hoechst to compare the results of different ECM coatings on day 5 (Fig. 4AII). Fibronectin coated chips had few attached cells, with the majority of the membrane left unoccupied. The membrane coated with Collagen-I had comparatively more cells attached but still had large sections void of any cells. Matrigel coating had a significantly larger proportion of cells attached to the membrane with few empty spots. The coating with the highest cell coverage was the mixture of Matrigel and Col-I, similar to the findings of Humayun M., et al. Fig. 4AII illustrates the distribution of area coverage by cells in each membrane and ECM scenario tested. The most confluent monolayer can be observed in the combination of collagen and Matrigel on a PC membrane.

3.4. Testing the functionality of the lung-on-a-chip model
3.4.1. Live and dead staining
To confirm the cell viability, cells grown in the chips for 3, 5, and 7 days were stained with live and dead staining kit. The aim was to note the day for the cells to get fully confluent with minimal dead cells in the chip. The chip was then connected to a syringe pump to provide a constant stream of media in the lower channel, while the cells were cultured at the air interface in the upper channel by removing the media. The number of live cells increased gradually from day 3 and were confluent by day 5. On day 5, the media on the upper channel was removed to create an airliquid interface. The cells were viable and growing well on day 7 i.e., after 48 h at the air-liquid interface (Fig. 4C). Thus, we decided to perform further characterization of the cell layer lung-on-a-chip model

Fig. 4. Calu-3 adhesion in the chip: A) I. Cell adhesion to the membrane A-II Selection of coating ECM on the PC membrane: Once the PC membrane was identified as the suitable membrane, different ECM coating was used to identify the optimal ECM for cell attachment and proliferation. All cells were stained with Hoechst stain. (Scale: 100 μm) B) Cell viability: Live and dead staining of the cells in the chip on day 7 (Scale: 100 μm). 

on day 7. With optimized seeding density and proper ECM coating, the results showed that the chip was able to grow and maintain a viable cell population up to 2 weeks.

3.4.2. Mucus secretion
Airway mucus is an extracellular gel, which traps inspired toxins and carries them out of the lung through ciliary beating and coughing, making the lungs highly defensive to environmental harm (Fahy and Dickey, 2010; Knowles and Boucher, 2002). Calu-3 cells contain submucosal glands, which are a source of airway surface liquids such as mucus, making it an optimal cell line for assessing cellular functionality (Knowles and Boucher, 2002). Alcian blue staining enabled the detection of secreted mucus in Calu-3 cells (Haghi et al., 2010; Inglis et al., 1998). A film of mucus was seen on the surface of the Calu-3 cell layer after day 3 (Fig. 5A). The representative microscope images shown indicate an increase in mucus secretion over the 7 days of growth, as the blue staining became progressively darker, covering more areas of the membrane. The amount of mucus produced over several days, measured as an RGBB ratio (Fig. 5B), increased with time.

Here, mucus secretion is used as an indicator of a differentiated cell

Fig. 5. Mucus staining of Calu-3 in the chip A) Microscopic images of mucus staining of Calu-3 cells at I. Day 3 and II. Day 7 B) A plot of RGBB ratio across several days of Calu-3 grown in the chip (Mean SD, n ¼ 8) (Scale: 100 μm). J. Shrestha et al. Organs-on-a-Chip 1 (2019) 100001 7

layer to goblet cells (mucus-producing cells) and, therefore, by extension, the physiological functionality of cells (Haghi et al., 2010). The RGBB value is an indirect measure of mucus production and shows variations in mucosal secretion, which enables a comparison of mucus secretion across several days. Since the secretion increased with time, it can be concluded that the cell layer grown in the chip represented physiological functionality.

3.4.3. Sodium fluorescein permeation
The permeability of the cell monolayer was tested by studying the permeation of sodium fluorescein (flu-Na) across the cell monolayer. There was a significant decrease in the permeability values in the cultured devices as compared to the blank devices (Fig. 6A). The significant difference in flu-Na concentration indicates a barrier formation by the cell junctions in the monolayer of cells growing on the membrane. This ability of the model to culture cells at an air-liquid interface and develop a tight barrier against flu-Na permeation, is another physiological function of respiratory epithelial cells.

3.4.4. Cell surface P-gp expression
Transport proteins are integral transmembrane proteins involved in the pharmacokinetics of many drugs (Haghi et al., 2010). They play a crucial role in the absorption of the drugs, its distribution, and elimination of waste metabolites; as such they are vital in maintaining the pharmacological barrier integrity throughout the body. P-gp is a plasma membrane glycoprotein, magnesium (Mg2þ)-dependent ATPase, and is a predominant drug transporter protein (Bebawy et al., 2001). P-gp is physiologically expressed on the surface of respiratory epithelial cells (Madlova et al., 2009). Commonly used drugs for treating respiratory diseases have shown interaction with P-gp, making it relevant for respiratory drug transport studies (Forbes and Ehrhardt, 2005). The P-gp expression of the cultured Calu-3 cells in the chips was analyzed using flow cytometry by direct immunolabelling. In comparison with the isotype control, chip grown cells displayed P-gp expression, which increased gradually over time (Fig. 6B). Nevertheless, this result further validates the functionality of Calu-3 cells in our lung-on-a-chip model and favours the suitability of the model to be used for in vitro pulmonary drug transport studies.

3.5. Effects of CSE on IL-6 and IL-8 production from cultured Calu-3 cells
Cigarette smoke contains over 4000 individual chemicals, which are noxious and carcinogenic (Burns, 1991). These chemicals, once deposited on the surface of the airways, absorb across the alveolar-capillary membrane to be circulated into the blood. The inflammatory response contributes to the structural and functional changes in the lung, leading to their destruction and subsequently the development of lung diseases. Cigarette smokers are associated with altered levels of inflammatory cytokines secretion in their bronchoalveolar lavage (BAL) (Maestrelli et al., 2001; Mikuniya et al., 1999). IL-8 is a potent chemoattractant for neutrophils and eosinophils; released by phagocytes and various tissue cells when exposed to inflammatory stimuli (Baggiolini and Clark-Lewis, 1992). IL-6 is a pro-inflammatory cytokine secreted by epithelial cells and macrophages in the airways (Mikuniya et al., 1999). Several studies have reported increased levels of IL-6 and IL- 8 in the BAL and induced sputum obtained from smokers (Mikuniya et al., 1999; Mio et al., 1997). Budesonide, a glucocorticoid, has been successfully used in asthma and inflammatory disorders like COPD (Lung and Institute, 2002; Keatings et al., 1997). The cytotoxicity assay showed that 10% CSE was not toxic to the cells (Supplementary Materials). Thus, 10% CSE was used for the

Fig. 6. A) Flu-Na permeation: Transport of flu-Na from apical to basolateral in blank and seeded chips on day 7 of culture (n ¼ 3, * ¼ p < 0.01) B) Comparison of the P-gp expression in Calu-3 cells cultured in the chip on different days: The expression of P-gp increased with days. C) Flow cytometry results of measuring P-gp expression by Calu-3 cells on day 7: I. Control group with no staining; II. Control group stained with FITC-anti P-gp III. Control group stained with PI IV. Samples stained with both FITC-anti P-gp and PI.

experiments. Significantly increased secretion of IL-6 (p < 0.01) and IL-8 (p < 0.05) was observed with CSE treatment. The cells treated with 10% CSE only released significantly higher amounts of IL-6 (p < 0.01) and IL-8 (p < 0.05) over 24 h compared to cells treated with both CSE and Budesonide (Fig. 7A). The CSE-Bud and Bud-CSE group had lower amounts of IL-6 and IL-8 secretion, indicative of the anti-inflammatory effects of Budesonide (Keatings et al., 1997; Confalonieri et al., 1998). The increased secretion of IL-6 and IL-8 on exposure to CSE are in substantial agreement with previous studies (Mikuniya et al., 1999; Mio et al., 1997; Wang et al., 2000). The capability of our model to successfully replicate the effects of CSE and Budesonide makes it a suitable in vitro model for toxicological and inflammatory studies.

3.6. Effects of CSE on cellular expression of E-cadherin
The junctional complexes, tight junctions, and adherens junctions maintain the epithelial cell integrity (Jang et al., 2002). The tight junctions are located apically and form a barrier controlling paracellular permeability of ions and solutes, while the adherens junctions hold the cells together through calcium-dependent adhesion molecules (Tsukita et al., 2001; Zihni et al., 2016). E-cadherin, an adherent junction protein, is observed in the cell membrane or cell-cell junctions. It is required for proper localization of essential tight junctional proteins like claudin-1 and 4, and ZO-1, into junctions to maintain a correct and intact tight junction formation (Tunggal et al., 2005). The loss of this cell transmembrane protein destabilizes the epithelial cell integrity, leading to dissociation between neighbouring cells and reduced its polarity (Willis and Borok, 2007). Cigarette smoke exposure has been associated with the disruption of the airway epithelium barrier integrity and increased permeability (Rusznak et al., 2000). Also, the disruption of the intercellular junctional integrity in the absence of tight junction proteins when exposed to CSE has been previously noted (Schamberger et al., 2014). To confirm the effects of CSE treatment on the expression of the surface protein, E-cadherin, the Calu-3 cells were stained with monoclonal antibodies against E-cadherin and observed with a confocal laser scanning microscope. Single confocal slices are shown in Fig. 7 B illustrate the differentiated tight junction and nucleus stained regions. The cells grown in absence of CSE maintained their morphology and strongly expressed E-cadherin protein. However, treatment with 10% CSE for 24 h, showed reduced expression of the epithelial marker. Similarly, the cells treated with both CSE and Budesonide also had a slightly reduced expression of the E-cadherin. The results agree with earlier findings of the effects of CSE on the barrier integrity and E-cadherin expression in cultured airway epithelial cells (Schamberger et al., 2014; Oldenburger et al., 2014; Xi et al., 2000). This re-affirms the validity of our microfluidic lung-on-a-chip model as a functional platform to conduct pulmonary studies.

4. Conclusion
A simple fabrication technique of lung-on-a-chip devices using surfacetreated 3D-printed molds has been proposed in this paper. The fabrication technique allows the chip to be fabricated in less than a day, and the molds can be used for repeated PDMS casting. Thus, the technique is simple, robust, and cost-effective. Validation of our design was conducted by investigating the effects of CSE on the production of inflammatory markers, IL-6, and IL-8, along with effects on cellular expression of adherent junction protein, E-cadherin. CSE augmented the release of IL-6 and IL-8 from the epithelial cells cultured within a lung-on-a-chip model. The budesonide treatment helped to reduce the effects of CSE on the release of both IL-6 and IL-8. Moreover, our results show that the CSE disrupted the airway epithelial barrier by affecting the junctional proteins. This result is supportive of the findings of numerous studies suggesting that cigarette smoke is the precursor of inflammatory response and

Fig. 7. Effects of CSE on cultured Calu-3 cells in the chip: A) IL-6 and IL-8 release from Calu-3 cells in response to CSE: The cells were cultured until confluency in the upper channel and were treated with 10% CSE. Normal media alone or media containing 100 nM Budesonide was flown at 30 μl/h and collected in a tube. After 24 h of incubation, the collected media was later assessed for IL-6 and IL-8 by ELISA. B) Changes in the E-cadherin expression: Imaging of immunofluorescence-stained Calu3 cells cultured in the absence of CSE showed strong surface expression of E-cadherin. The Bud-CSE group and CSE-Bud group both showed a slight reduction in expression of the epithelial marker, E-cadherin. However, with treatment with 10% CSE for 24 h, the expression of E-cadherin was reduced significantly. E-cadherin staining is shown in green, and DAPI staining is shown in blue. Scale bar: 50 μm. White arrows indicate disrupted tight junctions after treated with 10% CSE. Data are expressed as mean SD (n ¼ 3). * indicates p < 0.05, ** indicates p < 0.01. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) contributes to the development of smoke-related respiratory diseases. Therefore, these results confirm that our design produces similar results to the in vitro static experiments but in a dynamic condition which has far more comparability to the real physiological conditions. The model developed using 3D-printed molds was able to maintain excellent barrier integrity, expressed cell surface functional P-gp, and secretion of mucus layer, providing a platform for permeability assays, transport mechanisms, and pulmonary drug delivery studies. Also, the ability to rapidly prototype these molds with little technical skills makes organ-on-a-chip modelling accessible to a broad group of researchers.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments
We would like to acknowledge the Australian Research Council for supporting through Discovery Project Grants (Grant Nos. DP170103704 and DP180103003) and the National Health and Medical Research Council through the Career Development Fellowship (Grant No.) APP1143377. Dr. Ghadiri is the recipient of Ann Woolcock Fellowship from Woolcock Institute of Medical Research.

3D printing on glass for direct sensor integration

by

3D printing on glass for direct sensor integration

Lin Ding, Sajad Razavi Bazaz, Mahsa Asadniaye Fardjahromi, Flyn McKinnirey, Brian Saputro, Balarka Banerjee, Graham Vesey & Majid Ebrahimi Warkiani

Significant improvements are being made in 3D printed microfluidics. 3D printing of microfluidic prototypes gained importance due to the fabrication flexibility compared to conventional techniques. Applications using these devices often require optical access to internal channels but even clear resins create translucent channels due to surface roughness and imperfections. This paper describes a 3D printing approach to form fluidic channels directly onto glass substrates that allows optical access to fluidic channels without distortion from 3D printing material. The glass substrate is itself a part of the flow channel which guarantees optical transparency. Micro- fabricated conductivity and impedance spectroscopy sensors were fabricated on glass substrates and placed in a custom made 3D printer build plate before fluidic structures are directly printed on top of the sensors. The effects on sensor performance and properties were evaluated using co-linear four-point probe resistance measurements, Raman spectroscopy, and impedance spectroscopy. It was shown that no resin or other chemicals are left behind from the printing procedure and sensor performance was unaffected. A proof of concept impedance-conductivity sensor was integrated with a 3D printed flow channel and shown to work as both conductivity and bacterial cells detection sensor.

We kindly thank the researchers at Montana State University for this collaboration, and for sharing the results obtained with their system.

Introduction

Additive manufacturing (AM) is becoming an increasingly viable option for the creation of fluidic devices [1–6]. Vat polymerization printing (VPP) such as stereolithography (SLA) and digital light processing (DLP) are popular techniques for creating 3D printing microfluidics due to their high resolution [7, 8]. The resolution of various printing techniques such as VPP, material extrusion, and material jetting are compared in [1–3,5,7]. 3D printing allows one-step creation of fluidic devices with high throughput and straightforward device parameter adjustment in Computer-Aided Design (CAD) software. Designs and devices can be distributed and replicated in different facilities by merely sharing the design files [1]. Commercial availability of printers and resins capable of truly microfluidic (sub 100 μm features) devices is somewhat limited [9]. In this paper, we demonstrate the integration of micro-fabricated sensors on glass with a high-resolution 3D printing technique.

Optically transparent surfaces within microfluidic devices are essential for accurate quantification of chemical, biological, and mechanical interactions [10]. Many 3D printing resins, however, are not clear, and while the use of clear resins theoretically allow for the creation of transparent devices, inherent surface imperfections can cause light diffusion creating translucent channels. There are several examples in the literature of devices printed on glass to create a very smooth surface using SLA and DLP techniques. Urrios et al utilized a glass build plate and glass vat to decrease surface roughness and increase print transparency in bio-microfluidic devices [11]. Gong et al constructed a custom printer to create microfluidic devices on glass substrates in order to avoid using an anodized aluminum build plate and to provide optical access to internal microfluidic components [12–14]. Parker et al 3D printed microfluidic devices with immunoaffinity monoliths on glass allowing for fluorescence measurements for the extraction of preterm birth biomarkers [15]. Lee et al 3D printed Quake style valves using glass as a build plate to increase channel visibility, eliminating the need to remove the device from the build plate, and to provide a flat, stable surface for a microscope stage [16]. Kim et al printed multiple designed microchannels on glass for convenient optical access [17, 18]. Beauchamp et al characterized the 3D printing of sub-100 μm external and internal, positive and negative resolution features and additionally created a particle trapping device with the particles clearly seen as viewed through glass [19]. Beauchamp et al again printed on glass to create visible microfluidic channels and fluorescent emission of preterm birth biomarkers inside the channel was collected [20]. Rogers et al fabricated 3D printed microfluidic channels with valves, and claimed that their ability to print directly on glass opened the possibility for direct integration of printing on materials with patterned electrodes [21]. The previous examples of printing on glass utilized base layers to ensure proper attachment, limiting the possibility of direct interaction between sensor material and fluids. Rather than printing directly on glass, Plevniak et al used an SLA printer to print a 50 μm open channel and then covalently bonded to glass post- printing resulting in a closed hollow channel [22]. Notably, Kou et al printed microfluidic channels directly on glass without the use of base layers to create an optical window for phase-contrast microscopy and fluorescence microscopy. [23]. However, a direct sensor integration in the 3D printed fluidic channel was not shown.

Integrating sensors for 3D printed fluidics is reviewed by Li et al [24]. Banna et al embedded pH and conductivity sensors into a fused deposition modeling (FDM) 3D print post-print allowing for sensor removal and replacement [25]. Using a print- pause-print technique, Gaal et al demonstrated a process for directly integrating an FDM printed structure with an electronic sensor by printing directly on the sensor rather than inserting the sensor post-print. Gold interdigitated electrodes (IDEs) were evaporated on a transparency sheet and a polydimethylsiloxane (PLA) channel was printed directly onto the transparency sheet creating an integrated electronic tongue sensor with optical window for increased channel visibility [26]. Pol et al described a fully integrated screen-printed sulfide-selective sensor on a 3D printing potentiometric microfluidic platform accomplished by a screen-printing step between two main FDM printing stages [27]. Di Nova et al designed, fabricated and tested a fully aerosol jet printed (AJP) electrochemical microfluidic sensor. Dispensed and cured silver- silver chloride and carbon ink formed electrochemical sensing elements on an alumina substrate while UV-curable polymer ink created a flow channel surrounding the sensing elements [28]. While these techniques work well for sensor integration, none utilize a glass substrate for optically clear channels.

Glass substrates are a viable method for creating optically transparent channels using high resolution printing techniques such as VPP or material jetting described above, however sensor integration often requires a multi-step integration process. In this paper, a glass based impedance-conductivity sensor is directly integrated with a 3D printed flow channel as a proof of concept device. This device is shown to work as both a conductivity sensor and bacterial cells detector. This is accomplished by printing directly on a glass substrate with patterned electrodes without the use of base layers using a DLP resin printer. Build plate modification and glass silane treatment facilitated this direct integration. The motivation for optical access in flow channels is well documented above and further explored below. Resin-printer deficiencies were overcome and sensor performance and properties were quantified both before and after device printing. Finally, sensor functionality is demonstrated with various conductivity and bacterial cells solutions. This work is significant because the advancement and advantages of 3D printed microfluidics described above can be directly integrated with sensors to study biological processes.

2. Methods

2.1. Printer integration

A MiiCraft 50 was the 3D printer utilized in this work. This printer uses 405 nm light to selectively cure polymer to form 3D structures by area-projecting light using DLP as described in [7]. Clear BV007 resin (MiiCraft) was chosen for its advertised low viscosity, transparency, and printer compatibility

A build plate was manufactured with a 20 x 25 mm insert, shown in figure 1. Using a silicone gasket and a diaphragm vacuum pump (Masis, GZ35-12) a glass slide could be held in place during the printing process and quickly released afterwards. The glass slide protruded slightly from the plane of the build plate; this distance was measured and accounted for in the printer software. Mounting the glass flush with the edge of the insert made it possible to reference the edge of the glass in the software so that prints could be precisely aligned to the edges of the glass itself with better than 100 μm

Figure 1: Modified build plate with insert for glass slide and silicone gasket and vacuum hole to hold glass substrate in place during printing.

Results

Diced borosilicate glass (BSG) substrates (Borofloat 33, University Wafer), 20 mm x 25 mm x 700 μm, were submerged in a 2.0 vol% solution of 3-(trimethoxysily) propyl methacrylate in ethanol (Bind-Silane, GE Healthcare, 17-1330-01) for five minutes, then submerged in ethanol for five minutes, and finally held at 105◦C for five minutes. This silane treatment protocol has been used to couple polymerizing hydrogel to a glass surface [29], but also worked to couple the resin used in this work to the glass without using an initial base layer.

Upon printing completion, the print was submerged in a solution of Resin Away (Monocure PTY LTD) and sonicated for 1 minute, or until all uncured resin was removed. The print and substrate were rinsed with deionized water and post-cured in B9 Model Cure (B9Creations, USA) for 20 seconds. Lab grade isopropanol was found to be less effective than Resin Away, as it would often cause the print to crack.

Physical vapor deposition (PVD) with thermal evaporation was used to deposit 10 nm of Cr adhesion layer, followed by electron-beam evaporation to deposit a 100 nm layer of Au onto a 700 μm thick, 100 mm BSG wafer. A positive photolithography process and wet chemical etch defined the patterned sensor, seen in figure 2. The wafer was diced into nine 20 x 25 mm sensors. These sensors were designed originally as temperature and electrochemical impedance spectroscopy (EIS) sensors, however in this work the temperature function was not utilized.

A flow chart outlining the fabrication process is shown in figure 3(a). Due to the limitation of the resin used in this work, printing closed channels was not deemed feasible. Instead, channels were printed open and a second print was used as a cover. The open channel and cover were each printed on glass and alignment features were designed on the open channel. A #3 round brush was used to lightly coat the contact surface of the open channel with uncured resin and then the cover was aligned and held firmly in place with a clamp during a post-cure process which fused the two prints together, resulting in a well-defined closed channel seen in figure 3(b). While using two prints to create closed channels was possible and repeatable, using a resin capable of closed channels would still be preferred in order to create a higher precision, single-step device

Replaceable 3D printed barb style adapters allowed tubing (1 mm ID, 3 mm OD, Tygon R-3603) to be connected to the flow channel for fluid introduction. Double-sided Kapton tape was used as an intermediate layer between the channel inlet/outlet and the barb adapter. While this was enough to secure the adapter for a short period of time, magnets (R422-N52, K&J Magnets) were used to secure the adapters for longer intervals with higher bursting pressure. These replaceable adapters seen in figure 3(b) and figure 3(c) resembled those used by Atencia et al but rather than a blunt needle, a barb connector was printed [30]. One of the advantages of printing adapters is that it can be assured that the liquid from the tube never interacts with the surface of the magnet. 

The adapters were printed on glass without the use of base layers so that the sealing face of the adapters was smooth and consistent print to print. A 1 mm hollow punch was used to create a consistently sized hole in the tape and then the adapter was centered and pressed onto the tape. A 7 mm hollow punch was used to cut out the tape in a circle slightly larger than the adapter. When ready for use the protective sheet was removed from the backside of the tape and pressed onto the flow channel inlet/outlet. The tubing was attached to the barb adapter and a ring magnet fit around the tubing onto the adapter. A second, disc magnet (D42-N52, K&J Magnets) provided the clamping magnetic force from the backside of the substrate.

2.2. Component characterization

Light transmission through sample printed on glass was measured with a spectrophotometer (FastSpec 528, MicroLab, USA) to quantify optical clarity of printing on glass. The transmittance, defined as the ratio of light which passed through the sample to the incident light, was recorded in this manner. 502 nm and 660 nm light was emitted from an LED towards a light detector and the intensity was measured. Measurements were taken by placing a 3D printed sample between the LED and detector. One sample of each thickness was measured 5-10 times at these two wavelengths.

Figure 3: Fabrication process and image of device. (a) Flow chart of device fabrication. (b) Exploded view of designed sensor packaging. (c) Assembled device with tubing and magnets. 

To demonstrate the improved image clarity provided by printing directly on a glass substrate, 275 μm microbeads (REDPMS-1.080 250-300μm, Cospheric, USA) were placed in a suspension solution and directed through the integrated flow channel on the sensor under a microscope. Images were collected viewing the microbeads through both the glass substrate and through the printed flow channel.

The bursting pressure of the magnetic adapters was tested in a similar manner to [30]. A printed solid cylinder and adapter were attached with a piece of Kapton tape. The printed cylinder simulated the flow channel to which adapters were attached. For consistency, cylinders were printed on glass and peeled off and post-cured with the glass side being the side to which the tape was affixed. Tubing connected the adapter to a 10 mL Becton Dickson syringe placed in a Kent Scientific GenieTouch syringe pump. The adapter end of the tubing was submerged in water so that the appearance of bubbles marked the point at which the connection began to leak. The volume of air in the tubing and syringe was noted. Approximating air as an ideal gas, the ideal gas law could be rearranged into Equation 1

 P = PaVi/(Vi − dV )

where Pa is atmospheric pressure, Vi is the initial volume recorded, and dV is the volume dispensed by the syringe pump, respectively. The guaranteed pressure applied by the syringe pump for this specific syringe according to the manufacturer was approximately 6 atm. Atencia et al found a bursting pressure of approximately 5 atm using a similar procedure with Kapton tape and magnets [30].

2.3. Device characterization

Resistance measurements were performed on Au samples with identical deposition parameters in order to quantify the effect of the silane treatment on sensor performance. A custom-built co-linear four-point probe with 1.3 mm needle tip spacing and a Keithley 2450 SMU were used to conduct the measurements. Spherical spring-loaded testing tips with a diameter of 0.6 mm were used to guarantee a good electrical contact (uxcell P11J). Ten consecutive readings of each current polarity were acquired and averaged. This was repeated ten times for two samples before and after silane treatment.

In order to confirm no resin or other chemicals are left behind from the printing process or silane treatment, three sensors from the same process wafer were investigated with Raman spectroscopy. One sensor was left untreated, a second received the silane treatment, and the third received the silane treatment and the base of the flow channel was printed on the sensor. A spectra was collected from a spot both on the glass and the gold from each sensor with a Horiba Labram HR Evolution Raman spectrometer. Additionally, a scan of the cured resin was captured.

The impedance spectra of a conductivity standard (Biopharm 84 μS/cm) collected before and after printing, with and without magnets was obtained in order to confirm the sensor performance remained unchanged due to printing. Before applying the silane treatment and packaging the sensor, a flow channel with the same dimensions (not printed on glass) was printed and clamped on top of the sensor using a silicone gasket to seal the channel. Three spectra were obtained using a Hioki IM3536 LCR in the range of 1 kHz - 8 MHz, rinsing the flow channel with deionized water between measurements. Another three spectra were obtained with the magnets for the barb adapters in place. Three more spectra with and without magnets were obtained after printing was complete.

2.4. Sample applications

As a proof of concept application, various conductivity solutions were synthesized using conductivity standards (Biopharm) and deionized milli-Q water (GenPure xCAD Plus, Thermo Scientific, 18.2 MΩ cm). Solutions of 21, 42, 63, 84, 141 and 353 μS/cm were created using milli-Q water to dilute 84 and 1413 μS/cm conductivity standards with the assumption that the milli-Q water had an insignificant conductivity relative to the conductivity standards. Impedance spectra from 1 kHz - 8 MHz were collected in the same manner as described previously. A frequency of 100 kHz was chosen to calculate the conductivity. Using a linear model for the conductivity-temperature relationship shown in equation 2, a temperature coefficient α was calculated using equation 3

where κ and T were the measured conductance and temperature and κref and Tref were the advertised conductivity at the reference temperature of 25 ◦C, respectively.

As another proof of concept application, Escerichia coli K12 cells were suspended in milli-Q water and different dilutions were analyzed using impedance spectroscopy. A Petri dish was removed from refrigerator stock, and a single colony of E. coli K12 was scraped from a Petri dish, and introduced into a 50 mL Falcon tube containing 25 mL of 1X tryptic soy broth (TSB; BD Bacto). The Falcon tube was inserted into an incubator (37 ◦C) with orbital shaker (150 RPM) for 19 Hours. Subsequently, the Falcon tube was centrifuged at 4700 RPM for 10 min. TSB media was replaced from the Falcon tube with 25 mL of autoclaved milli-Q water and thoroughly mixed. Nine 1.5 mL micro-centrifuge tubes were arranged in linear sequence for dilution series. 1 mL of media containing E. coli K12 was pipetted from undiluted Falcon tube into a first micro-centrifuge tube, and was thoroughly mixed. Next, 100 μL of media was pipetted from first mixture, and combined with 900 μL of milli-Q water into second micro-centrifuge tube creating a 1:101 dilution of E. coli K12 cells to milli-Q water. This procedure was repeated until a dilution of 1:108 (ninth micro-centrifuge tube) from the original was reached. All steps were performed inside a biosafety cabinet to ensure sterile conditions.

To obtain cell counts, 100 μL of the final four dilution mixtures were pipetted onto Petri dishes containing tryptic soy agar (TSA; BD Bacto) and spread using glass spreaders. The Petri dishes were sealed with parafilm and left for 24 hours at room temperature to incubate. Cell concentrations were performed by counting the individual colony-forming units (CFU) established on each plate, multiplying by the level of dilution, and dividing by the amount of media pipetted.

Yang et al performed a similar experiment where Salmonella cells were suspended both water and phosphate-buffered saline (PBS) and the impedance response over a range of frequencies were recorded using interdigitated microelectrodes. It was found that cell concentrations could be distinguished down to 106 CFU/mL and higher in water, while cells suspended in PBS could not be distinguished [31].

3. Results and discussion

3.1. Component characterization

The transmittance of a blank glass slide yielded a value within 1 % of the theoretical transmission of approximately 92 % [32]. Parts printed directly onto glass resulted in transmission values less than that of glass, as seen in figure 4, but still averaged higher than 90 %. These data suggest that the 3D prints are not significantly scattering or absorbing the incident light, as the average transmittance is still within 5 % to that of glass.

The packaged sensor can be seen in figure 3(b) and again in figure 5(a) with the barb adapters and yellow Kapton tape with magnets and tubing attached. Figure 5(b) shows how the relative position of the microbeads to each finger of the impedance sensor and channel wall is well-defined. In addition, the bead diameter is more easily determined which may be of interest with objects of unknown size. Note: The diameter of the used beads has an advertised variation of 50 μm and our measurements with a calibrated optical microscope were within this tolerance. In contrast, in figure 5(c) the relative microbead position and diameters are more challenging to determine.

During testing of the magnetic connectors, the syringe pump compressed the syringe past the advertised guaranteed force of the syringe pump with one or two exceptions. It’s possible the syringe pump is capable of applying more force than advertised, or approximating air as an ideal gas breaks down below 6 atm. In any case, the syringe pump was unable to cause the magnetic adapters to leak by nearly fully compressing 10 mL of air. This was deemed more than adequate for this work.

3.2. Device characterization

After performing the resistance measurements on two gold samples before and after silane treatment, a paired t-test with a double tail p-value of 0.048 and 0.740 was found for sample 1 (pre- and post-treatments) and sample 2 respectively, using the set of ten means. A boxplot showing these data is shown in figure 6(a). The p-values suggest the silane treatment has minimal effect on the resistance of Au used for the sensors in this work. 

The Raman spectra are shown in figure 6(b) with the cured resin spectra shown first followed by the glass and gold surfaces pre- and post-silane treatment and post-printing. All gold and glass substrates look alike regardless of silane treatment or printing and none show the peaks of the cured resin around 3000 cm−1 . This suggests that no resin or other chemicals are left on surfaces.

Impedance spectroscopy measurements were performed to evaluate sensor performance before and after the printing on the sensor. These data are plotted in a Nyquist diagram, which shows the real resistance versus the negative imaginary resistance. The curves of the Nyquist plot overlap in figure 6(c) suggesting that sensor performance is negligibly affected by the printing process.

3.3. Sample applications

The six conductivity solutions can be clearly distinguished in a Nyquist plot in figure 7(a). The experimental cell constant of 0.0832 cm−1 was found by comparing a commercial conductivity meter (Omega, CDS107). Using the linear temperature conductivity model described in equation 2 and equation 3, the conductivity of the various solutions was calculated and plotted in figure 7(b). The error bars show the min-max uncertainty due to temperature uncertainty. Also plotted is the measured conductivity by the commercial meter. The sensor measured conductivity values within

8 % of the meter with the exception of the lowest and highest conductivity solutions which each varied by more than 12 % from the meter. The working range of the sensor is affected by both the cell constant and the linear temperature compensation model.

2.4 x 107 and 2.4 x 108 CFU/mL) of suspended E. coli K12 cells and the Nyquist curves are shown in figure 8. The three dilutions are clearly distinguished in this plot and show that this packaged sensor is capable of distinguishing concentrations of bacterial cells down to 2.4 x 106 CFU/mL. It was found that lower concentrations of cells could not be distinguished, therefore the detection limit for this sensor is 2.4 x 106 CFU/mL for E. coli K12 cells suspended in water.

3.4. Discussion

The described procedure demonstrates how printing on glass can be applied to an arbitrary micro-machined sensor for integrating 3D printing and sensors. The sample applications were chosen to suit the sensors, however similar micro-fabricated sensors could be designed for many other applications. The procedure of build plate modification and silane treatment are compatible with the higher resolution printers and resins described above, and therefore recent advancements of 3D printed microfluidics can be integrated with micro-fabricated sensors in this described manner..

4. Conclusion

In this work, a procedure for direct integration of glass substrate sensing platforms with complex and precise packaging is outlined. The glass substrate is itself a part of the channel and guarantees ideal optical access. Resistance measurements, Raman spectroscopy, and impedance spectra show that sensor properties are unaffected and no resin or chemicals are left on the sensing surface after printing on the sensor. Viability of materials other than Au could be tested for other applications with similar techniques. 

Microbeads were imaged through the glass substrate to show the improved optical properties compared to imaging through a thin printed channel wall using clear resin. As proof of concept applications, a simple impedance-conductivity micro-machined sensor was directly integrated with a 3D printed flow channel with a well-defined sensing area and was shown to work as a conductivity sensor and E. coli K12 dilution detector with a detection limit of 2.4 x 106 CFU/mL. Applying the same methods for device realization, any number of devices can be integrating with a 3D printed structure. The limitations on device design generally are associated with the 3D printer, its resolution and compatible resins. These issues will become less prevalent as the technology in this rapidly advancing field improves and becomes available. Overall, this process will benefit from all advances in 3D printing technologies, becoming an increasingly viable avenue for sensor-packaging integration for bio-sensing.

Acknowledgments

This work was performed in part at the Montana Nanotechnology Facility, a member of the National Nanotechnology Coordinated Infrastructure (NNCI), which is supported by the National Science Foundation (Grant# ECCS-1542210) and funded by Montana State University (MSU). Transmission measurements were performed with the help of C. Bahn from the MSU Chemistry Department. B. Kincaid and G. Foster (MSU machine shop) helped manufacture the build plates. We thank the MSU C. Foreman lab for assistance with the E. coli K12 preparation.

Materials

M50

Learn More

Rapid Softlithography Using 3D-Printed Molds

by

Rapid Softlithography Using 3D-Printed Molds

by Sajad Razavi Bazaz, Navid Kashaninejad, Shohreh Azadi, Kamal Patel, Mohsen Asadnia, Dayong Jin and Majid Ebrahimi Warkiani

Abstract: Polydimethylsiloxane (PDMS) is a long-standing material of significant interest in microfluidics due to its unique features. As such, rapid prototyping of PDMS-based microchannels is of great interest. The most prevalent and conventional method for fabrication of PDMS-based microchips relies on softlithography, the main drawback of which is the preparation of a master mold, which is costly and time-consuming. To prevent the attachment of PDMS to the master mold, silanization is necessary, which can be detrimental for cellular studies. Additionally, using coating the mold with a cell-compatible surfactant adds extra preprocessing time. Recent advances in 3D printing have shown great promise in expediting microfabrication. Nevertheless, current 3D printing techniques are sub-optimal for PDMS softlithography. The feasibility of producing master molds suitable for rapid softlithography is demonstrated using a newly developed 3D-printing resin. Moreover, the utility of this technique is showcased for a number of widely used applications, such as concentration gradient generation, particle separation, cell culture (to show biocompatibility of the process), and fluid mixing. This can open new opportunities for biologists and scientists with minimum knowledge of microfabrication to build functional microfluidic devices for their basic and applied research.

Keywords: 3D-printed molds, 3D-printing, microfluidic resin, microfluidics, soft lithography

We kindly thank the researchers at the University of Technology Sydney and Macquarie University for this collaboration, and for sharing the results obtained with their system.

1. Introduction

In recent years, there has been a new surge of interest in 3D printing, which is defined as building successive layers of materials to form a desired object.[1,2] The interest in 3D printing methods is twofold. First, the advent of 3D printing has triggered the creation of numerous intricate designs, whether in micro or macro scale, often implausible through conventional fabrication methods. Second, 3D printing enables quick evaluation of ideated solutions, often within the same day. Feature-wise selection of printing parameters and multistep printing processes enable users to pay extra attention to the tiny details of their objects.[3] In addition, material specifications (e.g., Young modulus or transparency) can be adjusted based on the printing method. It is estimated that the market size of 3D printing will triple in the next half-decade, growing from 7.3 billion dollars in 2017 to 23 billion dollars by 2022.[4] As structures manufactured by 3D printing methods can be in the range of micrometers to centimeters, a new challenge emerges for microfabrication.[5]

The miniaturization of high-cost, resource demanding, and time-consuming lab processes into a high-efficient, multifunctionalized, and integrated microchip has been considered as a revolution across many fields of science.[6] Microfluidics, the commercial name for this revolution, is ubiquitous in fluid mechanics, reagent mixture, cell biology, particle and cell separation, metabolomics and proteomics, forensic, and genetic analysis.[7,8] Microfluidic devices enjoy the proficiency of low reagent consumption, parallelization, portability, integrated several biological assays, small footprint, accurate measurement, and live feedback.[9] Compared to macroscale fluid handling, microfluidics provides end-users with an economical and ready-to-use microchip with faster reaction time and prompt analysis.[10,11]

There is a growing body of literature that recognizes the significance of lithography in the fabrication of PDMS-based microchannels. However, lithography is limited in its ability to fabricate nonstraight microchannels. For instance, for vascular behavior imitation, fabrication of 3D complex vessel branches is mandatory.[12] Moreover, there are major limitations in the fabrication of angular designs, such as a microchannel with a trapezoidal cross-section.[13] Furthermore, nonplanar structures as well as 2D and 3D nanolithography always introduce more complexity to the fabrication process.[14] In addition, advanced equipment and an adroit operator are essential for microfabrication processes, especially when the surface coating of the device is of interest.[15] For these reasons, research groups have tried to provide alternative methods for the fabrication of molds used in softlithography processes.[16] One such alternative is the use of 3D printing technology for the fabrication of softlithography molds. Among all 3D printing methods, stereolithography apparatus (SLA) and digital light processing (DLP) offer great advantages, making them ideal candidates for microfluidics and biomedical applications.[17] However, one of the limitations of 3D printed SLA/DLP master molds for softlithography is the requirement for tedious pretreatments prior to PDMS casting. The pretreatment of the resin is necessary to ensure the complete curing of the PDMS in contact with the resin. Otherwise, the surface of the PDMS replica in contact with the resin cannot be polymerized due to the presence of residual catalyst and monomers, and its transparency would be also compromised.[18] It has been observed that the effects of pretreating the master mold are more significant in channels with smaller feature sizes,[19] and, in the case of relatively larger 3D printed parts, this challenge is not significant.[20] To address this issue, many researchers have proposed various pretreatment protocols to treat the 3D printed master mold before PDMS casting.[18,19,21–24] As one of the first attempts, Comina et al. proposed to cover the 3D printed template with a specialized ink via airbrushing.[21] However, the effectiveness of that method depended largely on the thickness of the ink. Four procedures are commonly used among other proposed postprinted protocols: 1) UV curing; 2) surface cleaning (e.g., ethanol sonification and soaking); 3) preheating; 4) surface silanization. Waheed et al. introduced an efficient yet time-consuming pretreatment protocol for PDMS softlithography.[24] The postprocessing included a 5 min UV treatment followed by 6 h of soaking in an ethanol bath. Following the air plasma treatment for 1 min, the surface of the 3D printed template was silanized by perfluorooctyl triethoxysilane for 3 h.

Nevertheless, there is still no consensus about the optimum protocol for treating 3D printed templates for PDMS casting. In addition, the proposed protocols are time-consuming, laborintensive, and lacking reproducibility. Furthermore, the treatment parameters, such as UV curing time, preheating temperature, and duration, seem to be a function of the feature size; thus, differ from one experiment to another.[24] Also, preheating in particular is a common step in many procedures and often induces high levels of material strain, resulting in the formation of cracks within microstructures.[18,25] Most importantly, surface silanization of the 3D printed templates is essential to ensure the PDMS peels off, correctly. Some silanizing agents such as perfluorooctyl triethoxysilane are cytotoxic and are not suitable for biological applications. Thus, development of a resin suitable for master mold fabrication will reduce all these time-consuming steps.

To address the aforementioned issues, herein, we optimize the use of a new resin developed by Creative CADworks (CCW Master Mold for PDMS devices) (i.e., made of methacrylated oligomers and monomers) for the fabrication of master molds directly by the DLP 3D printing method. We show that the 3D printed templates obtained using this resin can be immediately casted with PDMS without any pretreatment or surface modification. By way of explanation, the process of master mold design to microchip fabrication has been reduced from a time frame of several days (for a conventional softlithography process) to less than 5 h. In order to showcase the functionality of this resin, four different microfluidic devices have been developed. Each device represents a specific application, including separation, micromixing, concentration gradient generation, and cell culturing. The surface of the PDMS replica obtained from the 3D printed mold is also evaluated to investigate the bonding quality of PDMS.

Apparatus Used

A Bottle of the Master Mold for PDMS Device Resin

Master Mold for PDMS

Learn More

The CADworks3D Ultra-Series Microfluidic 3D Printer

Ultra 50
3D Printer

Legacy

2. Results and Discussions

2.1. PDMS Characterization

It is well-known that the quality of the PDMS casted on the mold can affect the whole performance of the microfluidic device.[26] Hence, its quality must be analyzed before use. After fabricating the 3D printed molds and removal of any residual resin, PDMS was casted on the master molds. For the sake of comparison, two different molds were fabricated, one with a conventional DLP resin and the other with the newly developed microfluidic resin. The main challenge with conventional DLP resin is that due to the presence of unreacted monomers, complete polymerization of PDMS cannot occur, resulting in the presence of residual material on both the PDMS and the mold, as shown in Figure 1A. The comparison between the mold fabricated via conventional resin and the microfluidic resin reveals that these two molds have identical surface roughness, and the smallest channel height for the fabrication of molds can be achieved with a thickness layer of 30 µm. However, for this thickness layer, the curing time of each layer for the newly developed resin is 6.5 s, which is more than the conventional one which is 1.3–1.5 s; as more time must be devoted to the methacrylated resins to be completely polymerized and cured. All in all, the fabrication time for both molds took less than an hour which is much faster than other methods. Also, the inset in Figure 2A shows the contact angle of the 3D printed molds. The contact angle measurement reveals that both surfaces are hydrophilic; however, the microfluidic resin is slightly more hydrophilic than the conventional one.

By substituting the acrylate components with methacrylated monomers and oligomers (Figure 1B), we are able to create a clean temporary binding site between the PDMS and the 3D printed master mold. To demonstrate this, we applied both of the conventional DLP resin and the newly proposed microfluidic resin to a single design and investigated the boding properties of PDMS. Both molds were subjected to the same experimental procedure.

Figure 1. A) Schematic illustration of how acrylated DLP resins impact the surface finish of casted PDMS pieces. Residual catalysts and monomers present at the interface between the resin and PDMS impede the polymerization of PDMS components, leaving behind residual material. B) Demonstrating the improved performance of methacrylated resin over conventional acrylates in providing a smooth surface finish with no residual material. This is due to a lack of unreacted monomers and oligomers impeding the complete polymerization of PDMS.
Figure 1. A) Schematic illustration of how acrylated DLP resins impact the surface finish of casted PDMS pieces. Residual catalysts and monomers present at the interface between the resin and PDMS impede the polymerization of PDMS components, leaving behind residual material. B) Demonstrating the improved performance of methacrylated resin over conventional acrylates in providing a smooth surface finish with no residual material. This is due to a lack of unreacted monomers and oligomers impeding the complete polymerization of PDMS.

It has been proven that in UV-cured systems, cracks developed as a result of shrinkage forces between and after curing.[27,28] In the methacrylated systems, this shrinkage has an inverse relation to the initial viscosity.[28,29] As the modified resin is more viscous than the conventional ones, the chance of cracks appearing and propagating reduced significantly during the curing process. As such, the mold fabricated via the microfluidic resin has better stability and a very smooth surface compared to those fabricated by conventional resin. As Figure 2A indicates, in the conventional DLP resin, PDMS surfaces in contact with the surface of the resin were not properly cured, and uncured PDMS layers remained on both surfaces. It can be clearly seen that the casted PDMS fails to adopt the pattern of the mold, thoroughly. In addition, during the detachment of PDMS from the mold, PDMS tends to stick to the resin, confirming that the surface of the conventional DLP resin is not appropriate for PDMS casting. By analyzing the materials constituting the conventional DLP resin, it is believed that this problem is related to the chemical composition of the resin. We hypothesized that the remaining catalyst and monomers on the surface of the printed mold disrupt the complete polymerization of a thin layer of PDMS in contact with the mold. This can be clearly seen upon the removal of the PDMS replica from the mold (Figure 2A). As such, the “acrylate group” in the resin’s chemistry is not a suitable choice for PDMS casting; this has urged different scientists to explore time-consuming strategies for the surface treatment of DLP printed molds. Through extensive research conducted by Creative CADworks, a new resin which contains methacrylated monomers and oligomers has been developed. Casted PDMS does not react with the methacrylated monomers because the surface of the mold is free of residual monomer units that may impede PDMS polymerization. As Figure 2B illustrates, once a blade cuts through the PDMS layer down to the mold, the PDMS replica detaches easily. The operation of each device and the quality of bonding were also analyzed for a wide range of flow rates (to check the simulations results of surface roughness and bonding quality, see Section 2.2) with the experimental setup shown in Figure 2C. The results, as shown in Figure 2D, confirmed that there was no leakage observed between flow rates ranging from 0.1 to 5 mL min−1, which indicates that the proposed method for fabricating PDMS-based microdevice is an ideal candidate for a variety of applications.

Figure 2. PDMS casting process in A) conventional DLP resin and B) microfluidic resin. The insets depict the contact angles on the surface of molds. In conventional resin, PDMS in touch with the surface of the mold cannot provide a temporary bonding, and the surface of the PDMS cannot replicate the pattern used in resin. In microfluidic resin, as soon as the blade reaches the surface of the mold, PDMS start to detach from the surface, and it can easily peel-off. The mold after PDMS casting in microfluidic resin clarifies that there is not any residual of PDMS on its surface, while in conventional DLP resin, residuals are on the surface. C) Experimental setup used in these series of experiments is illustrated. D) No leakage was seen during the experiments after bonding of PDMS by plasma surface treatment method.
Figure 2. PDMS casting process in A) conventional DLP resin and B) microfluidic resin. The insets depict the contact angles on the surface of molds. In conventional resin, PDMS in touch with the surface of the mold cannot provide a temporary bonding, and the surface of the PDMS cannot replicate the pattern used in resin. In microfluidic resin, as soon as the blade reaches the surface of the mold, PDMS start to detach from the surface, and it can easily peel-off. The mold after PDMS casting in microfluidic resin clarifies that there is not any residual of PDMS on its surface, while in conventional DLP resin, residuals are on the surface. C) Experimental setup used in these series of experiments is illustrated. D) No leakage was seen during the experiments after bonding of PDMS by plasma surface treatment method.

2.2. Simulation Studies of Surface Characterization

Here, the effects of surface roughness on the velocity and shear rate distribution along the length of microchannel were investigated through simulation study by COMSOL 5.3a. For a smooth surface, Sa was set as 0.3 µm, and for a rough surface, Sa was assigned to be 1 µm. Different flow rates of 0.1, 1, 1.7, and 3 mL min−1 were tested to investigate the shear rates present in the devices. Figure 3A shows velocity and shear rate distribution along the length of the smooth microchannel (Sa = 0.3 µm). The two insets (Figure 3AI,AII) depict shear rate distribution across the bottom surface of the microchannel at flow rates of 0.1 and 3 mL min−1; and by increasing the flow rate from 0.1 to 3 mL min−1, the order of the shear rate increased 100 times. Furthermore, the shear rate distribution illustrates that in the middle of the microchannel, due to the high shear rate, there is a higher probability for the quality of surface bonding to be disrupted than at the edge of the microchannel. Moreover, shear rate distribution 50 µm from the inlet was investigated at heights of 2, 5, 10, and 15 µm (half of the channel height) from the bottom surface for four flow rates of 0.1, 1, 1.7, and 3 mL min−1 (Figure 3AIII–AVI). The trend observed illustrates that the shear rate is focused halfway across the width of the channel at the height of 2 µm; as the height increases, the focus is drawn away from the center of the channel.

Figure 3. Velocity and shear rate distribution along the length of microchannel for A) Sa = 0.3 µm and B) Sa = 1 µm. Part I and II of each section (i.e., A and B) stand for the shear rate distribution at the bottom of the microchannel for velocity of 0.1 and 3 mL min−1 (black arrows are first principal curvature of surface). In the smooth channel, the peak of shear rate focuses at the center of the microchannel, where, in the other one, it relocates to the edges of microchannel. Shear shear distribution along the width of microchannel at 2, 5, 10, and 15 µm for velocities of 0.1, 1, 1.7, and 3 mL min−1 are illustrated by parts III to VI, respectively. It shows that in rough microchannel shear rate is uneven.
Figure 3. Velocity and shear rate distribution along the length of microchannel for A) Sa = 0.3 µm and B) Sa = 1 µm. Part I and II of each section (i.e., A and B) stand for the shear rate distribution at the bottom of the microchannel for velocity of 0.1 and 3 mL min−1 (black arrows are first principal curvature of surface). In the smooth channel, the peak of shear rate focuses at the center of the microchannel, where, in the other one, it relocates to the edges of microchannel. Shear shear distribution along the width of microchannel at 2, 5, 10, and 15 µm for velocities of 0.1, 1, 1.7, and 3 mL min−1 are illustrated by parts III to VI, respectively. It shows that in rough microchannel shear rate is uneven.

For the rough channel, although the applied flow rates were the same as the smooth channel, the shear rate distribution was noticeably greater. There is more variation in the bottom surface of the microchannel, (identified by the black arrow) when Sa = 1 µm compared to 0.3 µm. The bottom layer of the shear rate distribution also illustrates that the shear rate focuses more on the edges of the microchannel rather than in the middle (compared to the smooth surface). Thus, the probability of bonding disruption will be relocated to the edge of the channel instead of the middle of the channel. Figure 3BIII–BVI show the flow rates of 0.1, 1, 1.7, and 3 mL min−1 for Sa = 1 µm at a point 80 µm after the inlet. These figures demonstrate that the shear rate distribution is uneven along the width of the microchannel. Also, the shear rate values for Sa = 1 µm are higher than those for Sa = 0.3 µm for all heights and all magnitudes of velocity. Thus, surface roughness in microfluidic devices must be small enough so as to not impact upon the performance of the device, and the bonding quality as well as measurement performed within a microchannel were not influenced by the surface roughness of the microchannel.

2.3. Microfluidic Devices for Liquid Handling

Particle sorting and separation have become important processes within diagnostics and biological sample handling.[30] The unique properties of fluids at the microscale can be exploited to provide a perfect platform for handling fluid samples. For instance, fluid inertia is often used for focusing randomly dispersed particles into a particular location for the aim of collection or separation.[31,32] Spiral microchannels require relatively high flow rates which needs strong permanent bonding. In order to achieve strong bonding, the surface of PDMS layers must be ultrasmooth to facilitate plasma bonding of the PDMS and withstand the high shear stress generated by the input velocity.

Figure 4A shows the whole-chip layout of a spiral microchannel used in this study. Surface characterization depicts that the Sa is around 0.3 while Ra is approximately 0.2. As Ra is evaluated randomly in a line, it is reasonable that its value be less than that of Sa which covers the whole selected area. The function of the spiral microchip was examined with 15 µm fluorescent particles to verify the bonding and blocking of the microchannel. Flow rates from 0.5 to 3 mL min−1 (with an increment of 0.5 mL min−1) were tested to examine the bonding between the microchannel and its base, as shown in Figure 4B. It was illustrated that the flow behavior for these particles was the same as those reported in literature, where flow rates below 1.5 mL min−1 dispersed particles at the inner wall. However, at flow rates more than 1.7 mL min−1 , particles were focused at the outer wall and could be easily isolated for further use.[33]

Figure 4. A) Whole-chip bright-field image of the spiral microchip. Ra, Sa, and height profile are identified in the figure. B) Experimental observation of 15 µm fluorescent beads at various flow rates from 0.5 to 3 mL min−1 . C) Experimental observation of micromixer along the length of the microchannel with its corresponding values of Sa, Ra, and height profile. The values of Sa reveal that PDMS microchannels from microfluidic resin are proper fluidhandling applications.
Figure 4. A) Whole-chip bright-field image of the spiral microchip. Ra, Sa, and height profile are identified in the figure. B) Experimental observation of 15 µm fluorescent beads at various flow rates from 0.5 to 3 mL min−1 . C) Experimental observation of micromixer along the length of the microchannel with its corresponding values of Sa, Ra, and height profile. The values of Sa reveal that PDMS microchannels from microfluidic resin are proper fluidhandling applications.

Micromixers have become an essential tool in the preliminary stages of many lab-on-a-chip processes. Previously, by gaining the efficiency of proximity field nanopatterning and 3D nanolithography, Jeon et al. proposed a micromixer by implanting 3D nanostructures within the channel to enhance mixing efficiency, especially at low Re where diffusion mixing is dominant.[34] It has been proven that the combination of mixing units in micromixers improves the mixing efficiency.[35] As such, two different planar mixing units (without obstacles) were selected and connected to form a hybrid micromixer, as depicted in Figure 4C. The results of this micromixer design illustrated the efficient mixing of two fluids to give a high mixing efficiency suitable for many applications. Moreover, height profile of the channel is similar to the input CAD file. The values of Ra and Sa for this micromixer were measured to be 0.248 and 0.596, respectively. As the flow regime in microfluidic mixers usually exists at a Re of less than 100,[36] indicating laminar flow, the surface roughness does not adversely affect the function of the device.

Microfluidic devices can be integrated to act as modular components of a larger process. A decrease in the turnover time between designs as well as increased design flexibility makes 3D printing a perfect candidate for the future modularization of microfluidic devices.[37,38]

2.4. Biological Applications

In vitro cell culture platforms play a crucial role in cell biology, cancer research, regenerative medicine, pharmacy, and biotechnology. Although 2D cell culture in planar dishes is still widely used, this oversimplified model fails to mimic the actual cellular microenvironment. Alternatively, 3D cell culture platforms (mostly in the form of multicellular spheroids) are far more realistic models, which can better mimic in vivo responses.[39] However, these static 3D systems are still sub-optimal and lack many of the critical features essential to a complex tissue microenvironment. Additionally, these systems cannot precisely control the chemical and nutrient concentration gradients over time and space. Furthermore, the oxygen tension and shear stress experienced by the cells are different from in vivo conditions.[40] To address these shortcomings, microfluidic 2D and 3D cell culture platforms have emerged recently, progressing along with the rapid advances in microfabrication techniques.[41] Such platforms offer several advantages to engineering a physiologically relevant biomimetic tissue.

Here, we chose a pear-shaped microchamber similar to the design proposed by Chong et al.[42] The authors used the pearshaped design to minimize the shear stress during continuous perfusion. To fabricate the arrays of the microchambers, Liu et al. used standard dry etching on a silicon substrate followed by PDMS softlithography. The dimensions and characteristics of the 3D printed microchamber are shown in Figure 5A. The total printing time starting from the initial design to the final product took only 45 min. MCF-7 cells with a concentration of 106 cells mL−1 in culture media (Roswell Park Memorial Institute (RPMI) 1640 with 10% fetal bovine serum (FBS) and 1% streptomycin–penicillin) were introduced into PDMS microchamber. The device was incubated for 24 h at 37 °C with 5% CO2. To evaluate the cell viability in the PDMS microchamber, live/dead cell double staining was performed. As shown in Figure 5B,C, more than 98% of the cells remained viable in the microchamber 24 h after the initial cell seeding. This confirms that no cytotoxic residual material had been left on the PDMS from casting on the 3D printed resin. Also, in cell culture platforms, flow rates exist in the order of µL min−1,[43] and the values of Ra and Sa, as shown in Figure 5A, indicate that the device is functional within its flow regime. Therefore, it can be concluded that the newly developed resin for 3D printing master molds is suitable for cell culture applications and does not compromise cellular viability. Currently, lung-on-a-chip studies using 3D printed microfluidic resin molds are under investigation in our group; these studies demonstrate long-term cell viability (more than a week).

Figure 5. A) Whole-chip image of the cell culture device with its related Sa, Ra, and height profile. B) Live and C) dead images of the cells after 24 h incubation, which show that cell viability in these devices are noticeable, and total numbers of dead cells are rare. D) Concentration gradient profile of two food colors of red and green. The results reveal that the newly developed microfluidic resin is suitable for cell culture applications.
Figure 5. A) Whole-chip image of the cell culture device with its related Sa, Ra, and height profile. B) Live and C) dead images of the cells after 24 h incubation, which show that cell viability in these devices are noticeable, and total numbers of dead cells are rare. D) Concentration gradient profile of two food colors of red and green. The results reveal that the newly developed microfluidic resin is suitable for cell culture applications.

The gradient of biomolecules plays a crucial in controlling various biological activities, including cell proliferation, wound healing, and immune response. One of the most popular types of CGGs that produces discontinuous concentrations is the tree-like CGG. This type of CGG is based on the fact that one can divide and mix the flow through bifurcations and pressure differences downstream. This type of CGG is usually used for cancer cell cultures, as these CGGs transfer more oxygen and nutrients to cells as they develop a convective mass flux. Among various tree-like CGGs proposed in the literature, we chose the S-shaped CGG design developed by Hu et al.[44] The authors used micromilling to fabricate the CGG on a polymethylmethacrylate substrate. Here, we developed the same structure in PDMS using softlithography based master mold fabrication from our new microfluidic resin. Figure 4D shows the characteristics of the fabricated CGG. The device has two inlets and six outlets to produce six different concentration ranges. To examine the performance of the device, we used two colors of food dyes (please refer to the Supporting Information for dye preparation protocol). The concentration profile of the fabricated CGG is illustrated in Figure 5D, which is similar to those reported by the literature.[44] Since the velocity in CGG devices is small,[45] surface roughness cannot impose problems on the binding of PDMS. For printing of planar structures, 3D printing can be performed with higher slice thickness, as a result of which, printing time will be reduced.

In summary, the microfluidic resin for 3D printing is an ideal candidate for fabricating different bio-microfluidic devices and can replace all cost-intensive and time-consuming fabrication methods.

3. Conclusion

In this study, we introduced a microfluidic resin for direct fabrication of master molds for PDMS softlithography, which can substitute other time-consuming master mold fabrication methods. Conventionally, the main components of SLA/DLP resins are acrylated monomers and oligomers. These materials cannot provide a temporary attachment to PDMS without leaving uncured PDMS on the surface of the mold, indicating that the PDMS cannot replicate the mold pattern. In the proposed master mold microfluidic resin, methacrylated monomers and oligomers have been used to facilitate PDMS casting, the proof of which was illustrated by fabrication of four benchmark microfluidic devices, including separator, micromixer, cell culture device, and a concentration gradient generator. In addition, the effects of velocity and shear rate distribution on the total performance of the microfluidic device were investigated numerically. It was shown that the surface roughness has to be small enough so as to not create extra shear stress endangering PDMS bonding. As the fabricated devices were tested in wide ranges of Re, we showed that there was not any leakage in these microfluidic devices. The height profile also confirmed that there was not any major discrepancy between the CAD geometry and the fabricated part. The results of the spiral microchannel for flow rates from 0.5 to 3 mL min−1 illustrated that the behavior of particles in spiral microchannel was in line with those reported in the literature, and the microchip could withstand high flow rates. The characterization of the micromixer also demonstrated that the proposed microfluidic resin was able to fabricate microchannels with different geometries, and the mixing result was appealing so that two tested color dyes mixed completely. In the conventional softlithography process, silanization is necessary to prevent the attachment of PDMS to the master mold, which can be detrimental for cellular studies. The 3D printed mold obtained from the microfluidic resin proposed here does not require any silanization, and the cellular studies in the PDMS-based cell culture device confirmed the biocompatibility of the resin. The 3D printed CGG device produced a stable gradient profile, implying the application of such a versatile 3D printing technique for effective drug delivery. As PDMS-based microchannels are ubiquitous in microfluidic devices, the present study can be considered as a milestone in the microfluidic field which can reduce the brainstorming-to-production from a time frame of several days (including the time required for conventional master mold fabrication and post-treatment) to less than 5 h (with the new proposed microfluidic resin).

Apparatus Used

A Bottle of the Master Mold for PDMS Device Resin

Master Mold for PDMS

Learn More

The CADworks3D Ultra-Series Microfluidic 3D Printer

Ultra 50
3D Printer

Legacy

4. Experimental Section

Resin: As SLA/DLP printing process has risen in popularity, concern over its compatibility with PDMS is now an issue. The commercial resins used for DLP 3D printing of microfluidic devices were acquired from Creative CADworks company are BV-003 and BV-007 (manufactured by MiiCraft, Taiwan), which have been broadly used in the literature[46–48] (please refer to the Supporting Information for a detailed description of these two resins). However, these resins proved to be not effective for PDMS casting. As previously mentioned, although certain surface treatments for 3D printed molds (prior to PDMS casting) have been trialed, all are either time-consuming, nonreplicable, or not effective. These two resins are composed of acrylated monomers and oligomers. However, the required surface treatment for PDMS casting impedes their further applications in microfluidic devices. Thus, methacrylated monomers and oligomers were substituted to form a microfluidic resin, which is suitable for direct PDMS casting without any post-treatment. In conventional DLP resins, COCHCH2 exists in their functional groups. These components are not proper for the PDMS casting (i.e., incomplete cure of PDMS), and several groups tried to come up with a surface treatment strategy to mitigate this issue.[18] This problem is attributed to the acrylate groups, resulting in the utilization of methacrylated monomers and oligomers instead of them. Indeed, hydrogen (H) in the chemical formulation of acrylate components was replaced by methyl (CH3) to form the COCCH2CH3 group. The resultant resin possesses a viscosity in the range of 175–230 cps.

The polymer network of the methacrylate composites was shaped by the so-called process of “free-radical addition polymerization” of the corresponding methacrylate monomers. The process of polymerization happens in three stages, which are initiation, propagation, and termination. In this process, usually volume shrinkage is observed as a result of Van der Waals volume or the free volume reduction.[49] This volume reduction can be minimized by either adding the prepolymerized resins to the monomer resins, utilizing methacrylate monomers with high molecular mass, or increasing the percentage of inorganic filler. These monomers modify the final surface of the resin and eliminate the uncured layer in contact with the PDMS, making it appropriate for PDMS casting.[50] The exact formulations and chemical compositions of the developed microfluidic resin are proprietary to Creative CADworks.

3D Printer Specifications, Printing Parameters, and PDMS Casting: In this study, to create the molds, a MiiCraft Ultra 50 3D printer (MiiCraft, Hsinchu, Taiwan) was used, which has a printing area of 57 × 32 × 120 mm3 and XY resolution of 30 µm. The UV wavelength used in this device is 385–405 nm, which projects from the bottom of the resin bath filled with microfluidic resin. The operating temperature is 10 to 30 °C, and the operating humidity is 40% to 60%. The desired geometries were drawn in Solidworks 2016, a commercial CAD/CAE software, and then exported with the STL file format suitable for 3D printers. The STL file is imported into the Miicraft software (MiiCraft 125, Version 4.01, MiiCraft Inc), a software for preprocessing of design models. The imported file must be sliced to shape the desired geometry. The slicing in Z direction can be adjusted from 5 to 200 µm (with an increment of 5 µm). Reducing the thickness layer increased the final quality of the product. Since the modified resin has a high viscosity, the curing time of each layer is a challenging factor. In addition, the base and buffer layers must be carefully adjusted to allow the part to adhere to the picker without falling. When selecting a slice thickness of 10 µm for smaller features, it was better to set the curing time for each layer between 5 and 6 s. For slice thicknesses of 30 and 50 µm, the optimum curing times were found to be 6.5 and 9.5 s, respectively. The base layer is the layer that accounts for the bonding of the part to the picker. The curing time for the base layer was set to 60 s. The buffer layer was used to reduce the curing time between the base layer and subsequent part layers. As the UV light cures each layer, the Z-axis stepper motor displaced the sample one slice upward, before curing the next layer. This process continued until the whole geometry was printed. Once printed parts were removed from the picker, they were rinsed thoroughly with isopropanol. Next, an air nozzle was used to remove residual resin from the edges and in between extremely fine features. Eventually, the mold was postcured by exposing each part to the UV light in a curing chamber with a wavelength of 405 ± 5 nm. Upon fabrication of master molds, the PDMS prepolymer and the curing agent (Sylgard 184 from Dow Corning, MI, USA) were mixed in the ratio (W/W) of 10:1. This process was followed by degassing in a vacuum chamber for 15 min and pouring the liquid PDMS onto the 3D printed microfluidic mold without any surface treatment process. Afterward, it was kept in an oven to complete the curing of PDMS. Subsequently, the cured PDMS was peeled off, and the inlet and outlet holes were punched. The PDMS-based microchannel was then bonded onto either a glass slide or another PDMS substrate by plasma activation to form a closed channel. The schematic illustration of microchip fabrication based on the proposed resin is illustrated in Figure 6.

Figure 6. The workflow of the master mold preparation by DLP/SLA 3D printing method and microfluidic resin. A) The desired master mold is drawn. The beauty of microfluidic devices is that they require neither intricate geometries nor professional CAD drawer. Thus, the CAD drawing process will not take a long time. B) The design is then printed using a DLP/SLA 3D printer, and the residuals are removed from the surface of the mold. C) Afterward, PDMS is poured in the master mold, and D) in the final stage, PDMS is peeled-off, bonded to a glass or PDMS layer, and the finalization followed by the installation of inlets and outlets.
Figure 6. The workflow of the master mold preparation by DLP/SLA 3D printing method and microfluidic resin. A) The desired master mold is drawn. The beauty of microfluidic devices is that they require neither intricate geometries nor professional CAD drawer. Thus, the CAD drawing process will not take a long time. B) The design is then printed using a DLP/SLA 3D printer, and the residuals are removed from the surface of the mold. C) Afterward, PDMS is poured in the master mold, and D) in the final stage, PDMS is peeled-off, bonded to a glass or PDMS layer, and the finalization followed by the installation of inlets and outlets.

Benchmark Microfluidic Devices: In order to investigate the microchips fabricated via the 3D printed microfluidic mold, four benchmark
devices were tested. Generally, microfluidic devices are classified into two categories, those for liquid-handling and those for biological application.[51] To showcase liquid handling using the proposed 3D printing resin, a spiral microchip for separation and a micromixer for mixing two fluids were fabricated.

It has been shown that spiral microchannels with a trapezoidal cross-section are useful in particle/cell separation for a wide range of flow rates.[52] However, the fabrication of the mold which was mainly conducted by micromilling is a challenging process and not suitable for fabrication of complex cross-sections. By testing this device (please refer to the Supporting Information for sample preparation), the feasibility of fabricating a 3D-direct-printed spiral mold with a trapezoidal cross-section was evaluated, and the surface profile of the microchip and the bonding quality were assessed.

Mixing is an essential step in most chemical processes, and micromixer is an integral part of micro total analysis systems (µTAS). As such, the feasibility of producing planar micromixers has been showcased with a combination of two different mixing units adopted from Hossain and Kim[53] and Bhopte et al.[54] using the aforementioned technique (please refer to the Supporting Information for dye preparation). Finally, a specific design for cell culturing and concentration gradient generation for preparation of a drug with different dosages were selected. The cell culture device was selected to investigate the biocompatibility of 3D printed devices for cell culture applications (please refer to the Supporting Information for cell viability assay). The schematics of these devices with their specific dimensions are drawn in Figure 7.

Figure 7. Schematic illustration of certain microfluidic devices. Generally, microfluidic devices are divided into two categories of liquid handling and biological applications. Four benchmark devices for A) particle/cell separation, B) a specific well for cell culture, C) sample mixing, and D) a concentration gradient generator with their related dimensions are selected and illustrated.
Figure 7. Schematic illustration of certain microfluidic devices. Generally, microfluidic devices are divided into two categories of liquid handling and biological applications. Four benchmark devices for A) particle/cell separation, B) a specific well for cell culture, C) sample mixing, and D) a concentration gradient generator with their related dimensions are selected and illustrated.

Surface Characterization: Surface characterizations of the 3D printed mold and PDMS were analyzed using 3D laser microscopy (Olympus LEXT OLS5000); to this end, an LMPLFLN 20× LEXT objective lens (Olympus) was selected. Arithmetic mean deviation (Ra), the arithmetic mean of absolute ordinate Z (x,y) documented along a sampling length, and arithmetical mean height (Sa), the arithmetic mean of the absolute ordinate Z (x,y) documented along an evaluation area were chosen to evaluate the surface characterization of the samples. In order to investigate the velocity profile and shear stress along the length of the microchannel with a rough-embedded surface, COMSOL Multiphysics 5.3a, a commercial software based on the finite element method, was used. By applying the parametric surface function within COMSOL, two different Sa values (0.3 (attributed to the measured surface roughness of the spiral microchannel) and 1 µm) were evaluated. To apply roughness on the bottom of the channel, Equation (1) was used

where x and y are spatial coordinates, N and M are spatial frequency resolutions. The spectral exponent is controlled by β, and g(m,n)
and ϕ(m,n) are zero mean Gaussian and uniform (in the interval between −π/2 and π/2) random functions, respectively. In this study, the values of M and N were set to 40, and β was set as 2. Thereafter, f(x,y) was scaled in the Z direction to get the desired value of surface roughness.[55] Based on Equation (2), to identify the surface roughness, the amplitude parameter of Sa was used

where the mean-plane area is identified by A. A microchannel with dimensions 400 × 50 × 30 µm3 was considered, and the rough surface was applied at the bottom of the channel. In the simulations, flow was considered to be steady-state, incompressible, and Newtonian with the same properties as water. Uniform velocity was applied to the inlets, zero static pressure was assigned to the outlet, and all other walls were considered to be no-slip boundary condition.

MORE ACADEMIC ARTICLES

Supplementary Materials

Download

References

  1. Yenilmez, F. Ghaderinezhad, S. Katebifar, M. Messina, A. Khademhosseini, S. Tasoglu, Biofabrication 2016, 8, 022001.
  2. B. C. Gross, J. L. Erkal, S. Y. Lockwood, C. Chen, D. M. Spence, Anal. Chem. 2014, 86, 3240.
  3. S. Waheed, J. M. Cabot, N. P. Macdonald, T. Lewis, R. M. Guijt, B. Paull, M. C. Breadmore, Lab Chip 2016, 16, 1993.
  4. I. Wagner, Spending on 3D printing worldwide in 2019 and 2022 (in billion U.S. dollars), https://www.statista.com/statistics/590113/ worldwide-market-for-3d-printing/ (accessed: August 2018).
  5. A. I. Shallan, P. Smejkal, M. Corban, R. M. Guijt, M. C. Breadmore, Anal. Chem. 2014, 86, 3124.
  6. C. M. B. Ho, S. H. Ng, K. H. H. Li, Y.-J. Yoon, Lab Chip 2015, 15, 3627.
  7. A. Lashkaripour, A. A. Mehrizi, M. Goharimanesh, M. Rasouli, S. R. Bazaz, J. Mech. Med. Biol. 2018,18, 1850002.
  8. A. Lashkaripour, C. Rodriguez, L. Ortiz, D. Densmore, Lab Chip 2019, 19, 1041.
  9. M. Mollajan, S. R. Bazaz, A. A. Mehrizi, J. Appl. Fluid Mech. 2018, 11, 21.
  10. M. Rasouli, A. A. Mehrizi, M. Goharimanesh, A. Lashkaripour, S. R. Bazaz, Chem. Eng. Process. 2018, 132, 175
  11. A. Lashkaripour, M. Goharimanesh, A. A. Mehrizi, D. Densmore, Microelectron. J. 2018, 78, 73.
  12. P. F. Costa, H. J. Albers, J. E. Linssen, H. H. Middelkamp, L. Van Der Hout, R. Passier, A. Van Den Berg, J. Malda, A. D. Van Der Meer, Lab Chip 2017, 17, 2785.
  13. T. Kwon, H. Prentice, J. De Oliveira, N. Madziva, M. E. Warkiani, J.-F. P. Hamel, J. Han, Sci. Rep. 2017, 7, 6703.
  14. J. Park, K. I. Kim, K. Kim, D. C. Kim, D. Cho, J. H. Lee, S. Jeon, Adv. Mater. 2015, 27, 8000.
  15. Z. Mahmoodi, J. Mohammadnejad, S. R. Bazaz, A. A. Mehrizi, M. A. Ghiass, M. Saidijam, R. Dinarvand, M. E. Warkiani, M. Soleimani, Drug Delivery Transl. Res. 2019, 9, 707.
  16. M. A. Raoufi, A. Mashhadian, H. Niazmand, M. Asadnia, A. Razmjou, M. E. Warkiani, Biomicrofluidics 2019, 13, 034103.
  17. N. P. Macdonald, J. M. Cabot, P. Smejkal, R. M. Guijt, B. Paull, M. C. Breadmore, Anal. Chem. 2017, 89, 3858.
  18. H. N. Chan, Y. Chen, Y. Shu, Y. Chen, Q. Tian, H. Wu, Microfluid. Nanofluid. 2015, 19, 9.
  19. T. Dinh, H.-P. Phan, N. Kashaninejad, T.-K. Nguyen, D. V. Dao, N.-T. Nguyen, Adv. Mater. Interfaces 2018, 5, 1800764.
  20. K.-i. Kamei, Y. Mashimo, Y. Koyama, C. Fockenberg, M. Nakashima, M. Nakajima, J. Li, Y. Chen, Biomed. Microdevices 2015, 17, 36.
  21. G. Comina, A. Suska, D. Filippini, Lab Chip 2014, 14, 424.
  22. B. Parker, R. Samanipour, A. Ahmadi, K. Kim, IET Micro Nano Lett. 2016, 11, 41.
  23. P. H. King, G. Jones, H. Morgan, M. R. R. de Planque, K.-P. Zauner, Lab Chip 2014, 14, 722.
  24. S. Waheed, J. M. Cabot, N. P. Macdonald, U. Kalsoom, S. Farajikhah, P. C. Innis, P. N. Nesterenko, T. W. Lewis, M. C. Breadmore, B. Paull, Sci. Rep. 2017, 7, 15109.
  25. D. Karalekas, A. Aggelopoulos, J. Mater. Process. Technol. 2003, 136, 146.
  26. X. Ye, H. Liu, Y. Ding, H. Li, B. Lu, Microelectron. Eng. 2009, 86, 310.
  27. V. S. Voet, T. Strating, G. H. Schnelting, P. Dijkstra, M. Tietema, J. Xu, A. J. Woortman, K. Loos, J. Jager, R. Folkersma, ACS Omega 2018, 3, 1403.
  28. C. Charton, V. Falk, P. Marchal, F. Pla, P. Colon, Dent. Mater. 2007, 23, 1447.
  29. A. Ellakwa, N. Cho, I. B. Lee, Dent. Mater. 2007, 23, 1229.
  30. A. Tay, A. Pavesi, S. R. Yazdi, C. T. Lim, M. E. Warkiani, Biotechnol. Adv. 2016, 34, 404.
  31. A. Kulasinghe, T. H. P. Tran, T. Blick, K. O’Byrne, E. W. Thompson, M. E. Warkiani, C. Nelson, L. Kenny, C. Punyadeera, Sci. Rep. 2017, 7, 42517.
  32. M. Rafeie, J. Zhang, M. Asadnia, W. Li, M. E. Warkiani, Lab Chip 2016, 16, 2791.
  33. M. E. Warkiani, G. Guan, K. B. Luan, W. C. Lee, A. A. Bhagat, P. K. Chaudhuri, D. S. Tan, W. T. Lim, S. C. Lee, P. C. Chen, C. T. Lim, J. Han, Lab Chip 2014, 14, 128.
  34. S. Jeon, V. Malyarchuk, J. O. White, J. A. Rogers, Nano Lett. 2005, 5, 1351.
  35. S. R. Bazaz, A. A. Mehrizi, S. Ghorbani, S. Vasilescu, M. Asadnia, M. E. Warkiani, RSC Adv. 2018, 8, 33103
  36. N.-T. Nguyen, Z. Wu, J. Micromech. Microeng. 2005, 15, R1.
  37. A. K. Au, N. Bhattacharjee, L. F. Horowitz, T. C. Chang, A. Folch, Lab Chip 2015, 15, 1934.
  38. N. Bhattacharjee, A. Urrios, S. Kang, A. Folch, Lab Chip 2016, 16, 1720.
  39. Z. Koledova, 3D Cell Culture: Methods and Protocols 2017, p. 1.
  40. N. Kashaninejad, M. J. A. Shiddiky, N.-T. Nguyen, Adv. Biosyst. 2018, 2, 1700197.
  41. N. Kashaninejad, M. R. Nikmaneshi, H. Moghadas, A. Kiyoumarsi Oskouei, M. Rismanian, M. Barisam, M. S. Saidi, B. Firoozabadi, Micromachines 2016, 7, 130.
  42. L. Chong, W. Lei, X. Zheng, L. Jingmin, D. Xiping, W. Qi, C. Li, J. Micromech. Microeng. 2012, 22, 065008.
  43. M. Ni, W. H. Tong, D. Choudhury, N. A. A. Rahim, C. Iliescu, H. Yu, Int. J. Mol. Sci. 2009, 10, 5411.
  44. Z. Hu, X. Chen, L. Wang, Chem. Eng. Technol. 2018, 41, 489.
  45. A. G. Toh, Z. Wang, C. Yang, N.-T. Nguyen, Microfluid. Nanofluid. 2014, 16, 1.
  46. ] Y.-J. Park, T. Yu, S.-J. Yim, D. You, D.-P. Kim, Lab Chip 2018, 18, 1250.
  47. E. Mattio, F. Robert-Peillard, L. Vassalo, C. Branger, A. Margaillan, C. Brach-Papa, J. Knoery, J.-L. Boudenne, B. Coulomb, Talanta 2018, 183, 201.
  48. C.-K. Su, W.-C. Chen, Microchim. Acta 2018, 185, 1.
  49. J. V. Crivello, E. Reichmanis, Chem. Mater. 2014, 26, 533.
  50. R. Sakaguchi, J. Powers, Craig’s Restorative Dental Materials, 13th ed., Mosby, Saint Louis 2012, p.161.
  51. N.-T. Nguyen, M. Hejazian, C. H. Ooi, N. Kashaninejad, Micromachines 2017, 8, 186.
  52. M. R. Condina, B. A. Dilmetz, S. R. Bazaz, J. Meneses, M. E. Warkiani, P. Hoffmann, Lab Chip 2019, 19, 1961.
  53. S. Hossain, K.-Y. Kim, Micromachines 2014, 5, 913.
  54. S. Bhopte, B. Sammakia, B. Murray, presented at 2010 12th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), Las Vegas, Nevada, USA June 2010.
  55. M. Villegas, Z. Cetinic, A. Shakeri, T. F. Didar, Anal. Chim. Acta 2018, 1000, 248.

Quantifying EpCAM heterogeneity of circulation-tumor-cells (CTCs) from small cell lung cancer (SCLC) patients

by

Quantifying EpCAM heterogeneity of circulation-tumor-cells (CTCs) from small cell lung cancer (SCLC) patients.

H.Sorotsky, M.Aparanthi, D.Z.Wang, F.McFadden, S.N.Popescu, R.M.Mohamadi, M.Pereira, J.Weiss, D.Patel, S.Majeed, M.Cabanero, A.G.Sacher, P.A.Bradbury, N.B.Leighl, F.A.Shepherd, M.S.Tsao, G.Lui, S.O.Kelly, B.H.Lok

Objectives: To investigate the effect of pazopanib on different CTCs subpopulations in patients with recurrent SCLC and evaluate their clinical relevance.

We kindly thank the researchers at University of Toronto for this collaboration, and for sharing the results obtained with their system.

Methods

Methods: Blood samples from 20 SCLC pts were processed through the MagRC platform. Magnetic nanoparticles conjugated with anti-EpCAM antibodies were incubated with whole blood samples then introduced into the MagRC device where CTCs are sorted by differently sized nickel micromagnets within microfluidic channels. Captured CTCs are ranked into 8 zones that correlate with EpCAM expression levels (zone 1 = highest to 8 = lowest). For 8 pts, all samples were processed at a 1mL/hr flow rate (fr), and for 12 pts, a 0.5mL/hr fr was also studied; 66% of all chips were processed at a 1ml/hr fr and 34% at a 0.5ml/hr fr. The average zone for each chip was compared to the flow rate, age, and stage (extensive-stage (ES) vs limited-stage (LS)). The differences were tested using the Wald test within the linear mixed effects model.

Materials

A Bottle of the Master Mold for PDMS Device Resin

Master Mold Resin

Learn More

Results

Before treatment, CTCs could be detected in 50% of patients by CellSearch; phenotypic characterization of CTCs demonstrated that 50%, 46.6% and 27.6% of patients had CD45-/TTF1+, CD45-/CD56+ and TTF-1+/CD56+ CTCs, respectively. Additionally, 59% of CTCs were TTF-1+/VEGFR2+ and 53% CK+/VEGFR2+. One pazopanib cycle resulted to a significant decrease of the number of CTCs (CellSearch: p=0.043) and CK+/VEGFR2+ cells (p=0.027). At the time of PD, both the total number of CTCs (p=0.027) and the number of the different subpopulations were significantly increased compared to post-1st cycle values; this increased CTCs number was associated with a significant increase of TTF-1+/VEGFR2+ (p=0.028) and CK+/VEGFR2+ CTCs (p=0.018). In multivariate analysis, only the number of CTCs as assessed by CellSearch after one treatment cycle was significantly associated with OS (HR: 0.21; p=0.005).

Conclusions: Pazopanib has a significant effect on different subpopulations of CTCs in patients with relapsed SCLC; the detection of VEGFR2+ CTCs during treatment could be a surrogate marker associated with resistance to pazopanib.

Keywords: CD56; CTCs; CellSearch; Immunofluorescence; Pazopanib; SCLC; TTF-1; VEGFR2.

2nd Summer School on Complex Fluid-Flows in Microfluidics

by

2nd Summer School on Complex Fluid-Flows in Microfluidics

Francisco J. Galindo-Rosales

The second edition of the “Summer School on ComplexFluid-Flows in Microfluidics” was held at the Faculty ofEngineering of the University of Porto, Portugal fromJuly 9– 13, 2018 sponsored by Anton Paar, Applied Sci-ences, BlackHole Lab, Elveflow, Formulaction, the Por-tuguese Society of Rheology, and Rheinforce (in alpha-betical order). The company Creative CADWorks kindlyprovided microfluidic connectors, chips and molds fab-ricated with its 3Dprinter. This 5-days course (6h/day)intended to provide cutting-edge knowledge on com-plex fluid-flow at microscale to those researchers work-ing on microfluidics, with complex fluids or a combina-tion of both.The first day of the summer school was fully dedicatedto  “Complex  fluids  and  Rheometry  at  Microscale”.Three  of  the  four  different  approaches  to  performrheometry of a fluid sample with a characteristic di-mension smaller than 1 mm were covered during thefirst day of the summer school: Manlio Tassieri (Univer-sity of Glasgow, UK) presented the different principlesand applications of passive and active microrheology,Jan Vermant (ETH Zürich, Switzerland) shared his ex-pertise on interfacial rhe o logy, Hubert Ranchon (For-mulaction, France) showed how to perform rheometryon a chip with their Fluidicam Rheo, and finally Francis-co J. Galindo-Rosales (CEFT/FEUP, Portugal) divided histime into two presentations, one focused on the differ-ent approaches for performing extensional rheometryon a chip, and another one focused on how to exploitthe non-linear behavior of complex fluids at microscalefor developing damping composites with optimal per-formance under impact loads.The second day was focused on “Fabrication tech-niques in Microfluidics”. Benjamin Sévénié (BlackHoleLab, France) showed how to fabricate microfluidic chipswithout a clean room, Vânia Silverio (INESC Microsys-tems  and  nanotechnologies,  Portugal)  talked  aboutfabrication  methods  for  precision  microfluidic  inter-faces for the development of microchannel integrateddevices, Paulo Freitas (International Iberian Nanotech-nology  Laboratory,  Portugal)  lectured  on  magneto -phoretic and size based modules for biosensor applica-tions in microfluidics, and finally Paulo Marques (INESCTEC, Portugal) explained how to fabricate OptofluidicDevices by Femtosecond Laser Direct Writing and Ma-chining.

We kindly thank the researchers at University of Porto for this collaboration, and for sharing the results obtained with their system.

Figure 1: Pictures at different moments of the course: a) P.C. Sousa and b) J.M. Miranda during their parallel lab-sessions onfluid-flow characterization in microfluidics, c) J.D. Araújo and d) C.B. Fernandes during their lab-sessions on numerical opti-mization and computational simulation using OpenFOAM, e) J. Vermant and f) M. Tassieri at the beginning of their lectures.

The third day was centered on how to perform “Flu-id-flow characterization in Microfluidics”. During themorning, Benjamin Sévénié (in representation of Elve-Flow,  France)  talked  about  dispensing  with  pressurepump and measuring with flow sensor, and Mónica S.N.Oliveira (Strathclyde University, UK) lectured about different experimental techniques for performing a Fluidflow characterization at the microscale. The afternoonwas dedicated to an experimental lab-session, whereJoão M. Miranda (CEFT/FEUP, Portugal) demonstratedhow to generate and characterize microfluidic drop gen-eration, Patrícia C. Sousa (International Iberian Nano -technology Laboratory, Portugal) showed how to mea-sure velocity profiles with micro-PIV, and Francisco J.Galindo-Rosales (CEFT/FEUP, Portugal) showed differ-ent components typically used in microfluidic experi-ments, such as pressure/syringe pumps, pressure sen-sors, tubing and connectors, etc.

On the fourth day Alexandre M. Afonso (CEFT/FEUP,Portugal) and João M. Nóbrega (IPC/University of Min-ho, Portugal) lectured during the morning session on“Computational modelling of complex fluid-flows atmicroscales”. The afternoon was fully dedicated to acomputational Lab-Session, coordinated by Célio B. Fer-nandes (IPC/University of Minho, Portugal) and Luís L.Ferrás  (IPC/I3N/University  of  Minho,  Portugal),  sup-ported by J.M. Nóbrega and A.M. Afonso, respectively.

The last day was entirely dedicated to “Numericaloptimization in Microfluidics”. The morning session wasdedicated to the lectures of Kristian E. Jensen (Comsol,Denmark), who talked about the basics concept and op-timization with Finite Element Methods, and Manuel A.Alves (CEFT/ FEUP, Portugal), who focused on the ap -plication of optimization tech ni ques with Finite VolumeMethods  to  the  development  of  extensional  rheo -meters on a chip. The afternoon session was fully dedi-cated to a Lab-Session on numerical optimization tech -niques,  which  was  coordinated  by  Kristian  E.  Jensen(Comsol, Denmark) and José Daniel Araújo (CEFT/FEUP,Portugal), again with the support of A.M. Afonso.

The content of the course covered the three classi-cal approaches, i.e. theoretical, experimental, and nu-merical to tackle scientific problems related with com-plex fluid-flows at microscale. A book on applied rheol-ogy [1], which was kindly provided by Anton Paar, wasdistributed  among  the  participants.  The  course  wasconceived and planned to be interactive and practical,thus 3 hours of lectures were provided during the morn-ing sessions; followed by a 3-hours slot for lab-sessionsduring the afternoon sessions in the microfluidic labo-ratories of the Transport Phenomena Research Centre(CEFT/FEUP) and solving some exercises in the comput-er laboratory (Figure 1). Thus, 70% of the time was ded-icated to lectures and 30% was dedicated to lab ses-sions. From the lecture’s time, it is worthy to highlightthat 32% was given to the sponsor companies to talkabout their latest developments for microfluidics appli-cations  (Figure  2).  The  course  gathered  13  lecturers,5 lab-session  demonstrators  and  25  participants,  allcoming from 14 different nationalities, what make theatmosphere very multicultural and also allowed theparticipants to enlarge their network of potential col-laborators.

Looking  forward  to  your  participation  at  the  3rdSummer School on Complex Fluid-Flows in Microfluidics!

Acknowledgement
F.J.  Galindo-Rosales  would  like  to  acknowledge  thefinancial  support  from  FCT,  COMPETE  and  FEDERthrough  grant  IF/00190/2013  and  project  IF/00190/2013/CP1160/CT0003.

Figure 2: Some statistics about the participants and speakers during the event.

Materials

A Bottle of the Master Mold for PDMS Device Resin

Master Mold Resin

Learn More

H Series

Learn More