Fabrication of a Micro/ Nanofluidic Platform Via Three- Axis Robotic Dispensing System

The purpose of this work is to introduce a new fabrication technique for creating a fluidic platform with embedded microor nanoscale channels. This new technique includes: (1) a three-axis robotic dispensing system for drawing micro/nanoscale suspended polymer fibers at prescribed locations, combined with (2) dry film resist photolithography, and (3) replica molding. This new technique provides flexibility and precise control of the microand nano-channel location with the ability to create multiple channels of varying sizes embedded in a single fluidic platform. These types of micro/nanofluidic platforms are attractive for numerous applications, such as the separation of biomolecules, cell transport, and transport across cell membranes via electroporation. The focus of this work is on the development of a fabrication technique for the creation of a nanoscale electroporation device. [DOI: 10.1115/1.4034611]


Introduction
A number of different methods for creating microfluidic devices have been previously developed including direct fabrication techniques (photolithography, including X-ray and e-beam; laser photoablation; and laser micromachining) and prototyping techniques, such as injection molding, hot embossing, and soft lithography [1][2][3].Of these techniques, laser micromachining, injection molding, and soft lithography have been gaining popularity as methods of choice due to their compatibility with polymeric materials [4].Laser micromachining has been widely employed in fabricating microfluidic devices due to its high accuracy and resolution, ultrashort timescale, and ultrahigh laser intensity coupled to the electronic system [3].Even though there are a number of advantages to laser micromachining over traditional methods, it is limited in cost, centralized labor, optical opacity, and difficulty in requiring highly specialized expertise and equipment [5,6].Compared to other micromachining methods, laboratory-based injection molding methods are advantageous due to low cost, ease of operation, and flexibility [4].On the other hand, injection molding is limited in the complexity of the molding equipment and fabrication of the molds [4].Using soft photolithography, simple microchannels can be generated via casting elastomeric materials on top of micropatterned molds [7].The advantages of soft lithography include easy to learn, straightforward to apply, and low-cost, while the disadvantages are complexity in fabricating multilayer structures due to alignment, limited to amorphous materials, and lack of temperature range control [8].E-beam lithography has also been used to create micro/nanochannels in SU-8 substrates [2]; however, this technique requires very expensive equipment, is a slow process, and is not highly repeatable.In sum, although the micro/nanofabrication techniques defined above are viable for fabricating micro/nanofluidic platforms, there are issues with repeatability, large-scale production, and difficulty with generating nanoscale structures buried inside a bulk device.Thus, new techniques need to be developed to increase production throughput and expound on the realization of a new cadre of micro/nanodevices and micro/nanofluidic platforms, such as micro-/nano-electroporation devices.
Over the past 30 years, electroporation of biological cells has gained widespread attention due to its broad applications in cellular analysis, cellular transport, and transfection of biomolecules for cell therapy.Electroporation requires an electrical field applied at a specified intensity and duration to create pores in the cell wall.Traditional electroporation processes are achieved by applying an electric field across the entire cell membrane, which causes unwanted stress on the cell and significantly reduces cell viability.Huang and Rubinsky [9] utilized traditional microfabrication techniques to create a micro-electroporation platform to determine the effect of applying a focused electric field over a microsized region of the cell.They hypothesized that a microsized pore would reduce stress on the cell and improve cell viability.In their study, genes were successfully transfected into the cells with relatively high efficiency while increasing cell viability rate, reducing cellular stress and lowering sample contamination compared to conventional bulk electroporation devices.Nonetheless, these microfluidic electroporation platforms were plagued with Joule heating effects, an inability to precisely control the dose of transfected biomolecules, and the overall cell viability was still significantly lower than normal cultured cells, which is the desired goal of these electroporation platforms.
More recently, Boukany et al. [10] fabricated electroporation devices with nanoscale channels using a DNA combing and imprinting method to investigate whether a nano-electroporation platform would enable achievement of the desired cell viability and precise control of the transfected biomolecules.These nanoelectroporation platforms successfully transfected biomolecules into the cell in a controllable fashion and significantly improved cell viability to levels comparable to normal cell culturing techniques.Although their fabrication technique successfully created the desired nanochannels, several issues occurred with repeatability in fabricating the nanoscale channels, precision in controlling the nanochannel location within the device, device yield, and inability to automate the process for large-scale or mass production.Thus, the purpose of this study is to explore the development of a method that can repeatedly and precisely produce submicron scale to nanoscale channels within a polydimethylsiloxane (PDMS)-based fluidic platform, which can be utilized to perform electroporation in cells.

Methods and Materials
The design for this cell electroporation device consists of two microchambers in juxtaposition to one another with a gap separation distance of 10 lm (Fig. 1).The micro/nanofluidic PDMS devices were batch fabricated with either a microchannel (channel design diameter ¼ 1 lm), submicrochannel (channel design diameter ¼ 500 nm), or nanochannel (channel design diameter ¼ 300 nm) separating the two microchambers.The general fabrication procedure first involved making a mold of the reverse structure of the electroporation device.The next step in the fabrication process consisted of creating a mold with two microchambers on a glass substrate using dry film resist photolithography (Fig. 2).Subsequently, a prescribed polymethyl methacrylate (PMMA) microfiber or nanofiber was drawn via a robotic dispensing system that translated a microneedle from the initiation position across the center tip of two microchambers with high spatial precision.Then, PDMS was poured onto the glass mold consisting of the two microchambers and covered the microfiber or nanofiber.The last step involved etching away the sacrificial fiber buried in the PDMS by placing the PDMS structure in a beaker filled with acetone solution and applying sonication.Dry Film Photolithography.A photomask consisting of three sets of two juxtaposed tip-to-tip microchamber convex structures was designed in L-Edit and fabricated via a Heidelberg DWL 66fs Laser Writer (Heidelberg Instruments Mikrotechnik GmbH, Heidelberg, Germany) in combination with traditional photolithography techniques.The glass substrates were first prepared using organic solvents to clean the surface of the substrates.Subsequently, the dry film resist (DFR, t ¼ 15 lm), negative contrast (DFRs, Ordyl SY 320, ElgaEurope, Milan, Italy, a solvent type permanent dry film for special microelectromechanical system applications) was cut to the size of the substrate.Characterization of the DFR photolithography process included: (1) optimizing exposure time, postexposure bake time and temperature, and development time; (2) developing the DFR transferring process to make it adhesive to substrates without any bubbles; and (3) identifying the most appropriate development solutions for completely removing the resist's residue after development.The optimized experimental parameters are shown in Table 1.The DFRs provided several advantages compared to liquid resists, such as ease Fig. 1 Schematic of the design for the micro/nanofluidic electroporation devices (Note: To make the details viewable, the dimensions are not to scale.)Transactions of the ASME of use during the fabrication process, good planarity, better vertically thick structures (as thick as 12.5 lm), low cost, and excellent adhesion to different substrates including, but not limited to glass, Si wafer, and Delrin as previously described by others [11][12][13].
The next fabrication step consisted of detaching the easy-toremove protective layer of the DFR with a razor blade and transferring the resist onto the substrate using a hand roller.The resist and substrate were placed on an 85 C hotplate and allowed to bake for 2 mins.Once the substrate cooled, it was transferred to a custom-made collimated ultraviolet (UV) light source and exposure system, consisting of a 10 W 365 nm LED (Fig. 3).The photomask was placed on top of the resist-coated substrate and exposed to the UV light for 4 mins.After the exposure was completed, the substrate was treated with a postexposure bake for 2 mins at 85 C on a hotplate.Once cooled, the top protection layer of film was peeled from the resist for development.The substrate was developed in a 10:1 volume ratio of xylene to IPA for 1 min to solvate away the unexposed structures.
Fiber Direct-Writing Process.The prescribed micro/nanofibers were drawn on the micropatterned substrate using a three-axis robotic dispensing system (model JR 2203N; Nordson, Westlake, OH), which included a dispensing valve, valve controller, three-axis positioning system (spatial resolution of 5 lm in the X, Y, and Z direction), USB-microscope, feedback-controlled heater, and a sealed enclosure (Fig. 4).The direct-writing process of the fibers with the three-axis robotic dispensing system has been previously described [14].However for completeness, the components will be briefly described here.JR C-POINTS software was utilized to control the dispensing valve, valve controller, positioning of the fibers' initiation and termination location, syringe needle travel velocity, vertical distance of the syringe needle from the substrate, and dispense time.The dispense time refers to the time the PMMA solution was pressurized in the needle before fiber drawing begins.The USBmicroscope aided in video recording the entire direct-write process to monitor droplet accumulation on the needle and determine fiber break-up time.The feedback-controlled heater and plastic enclosure provided a constant temperature environment to ensure a uniform evaporation rate of the polymer solvent and prevented environmental air flow/circulation over the workspace, which can cause fiber breakage during the writing process.
The direct-writing process for each fiber consisted of four main steps as illustrated in Fig. 5. Prior to performing the direct-write process, samples of PMMA (Sigma-Aldrich, St. Louis, MO; MW ¼ 996 g/mol) were prepared by dispersing the PMMA into chlorobenzene at weight concentrations varying from 19% to 27% in 1% increments.The solutions were placed in an ultrasonic bath for $4 to 5 h to completely dissolve the PMMA in chlorobenzene.The PMMA polymer solution was loaded into the syringe and subsequently pushed out of the needle tip by pressurized air.The syringe needle was positioned to a predefined spot on the substrate and lifted up 1 mm from the initial touching point, and the syringe needle was laterally translated with a constant traveling velocity to the termination location on the other side of substrate.During this step, the PMMA solution droplet thinned and elongated by surface tension-driven necking as defined by the capillary drying process [15,16].The mathematical model indicating the direct  relationship between the viscosity and the surface tension of the polymer solution and break-up time is given as follows [17]: where D(t) is the diameter of polymer solution fiber as a function of time t, D 1 is the initial diameter of the fiber, X is a constant equal to 0.7127, r is the surface tension of the polymer solution, and g is the Newtonian viscosity.Although this mathematical equation illustrated a relationship between filament break-up time and the polymer solutions' viscosity and surface tension, it is limited to just a linear relationship description among those factors and does not provide an accurate prediction when the final filament diameter is zero, i.e., fiber break-up.Thus, Tripathi and coworker [16] developed a model for Newtonian fluids that incorporates evaporation rate in the model, where the equilibrium diameter, D 1 , after solidification and thinning without any breaking was shown to be where P is a dimensionless processability parameter, and P ¼ ðgv=rÞ, where v is the evaporation rate (solvent mass transfer coefficient) [18].Finally, the single fiber direct-write process was completed by lowering the syringe needle down 1 mm onto the terminating position of substrate.
Replica Molding.Once the micro-or nano-scale fibers were written onto the mold, the whole mold was placed in a Petri dish for replica molding.First, polydimethylsiloxane (PDMS) and its curing agent (1:10 weight ratio to PDMS) were mixed with a stirring bar for 2-3 mins.Subsequently, the PDMS mixture was poured slowly into the Petri dish covering the substrate mold taking great care to avoid breaking the suspended micro/nanoscale polymer fibers.Subsequently, the Petri dish containing the mold and PDMS mixture was placed in a vacuum chamber for 10 mins to remove air bubbles from the PDMS solution.Next, this Petri dish was transferred to an oven and cured at 60 C overnight.After curing overnight, the PDMS was peeled off from the glass substrate, and the embedded polymer fibers were removed by placing the PDMS substrate in an ultrasonic acetone bath for Fig. 4 Image of the three-axis robotic dispensing system placed in a temperature-controlled environmental chamber.The inset is an enlarged image of the microscope and pressurized injection system.Adapted from Ref. [14].30 mins.Due to the considerably high swelling rate of PDMS in acetone, the sonication time was carefully controlled to minimize the amount of swelling in order to maintain the integrity of the nanochannel.Afterward, the PDMS replica mold was exposed to oxygen plasma to make the two microchambers' and submicro/nanochannel's surfaces hydrophilic.For storage, the samples were submerged in methanol solutions or de-ionized water to maintain their hydrophilic properties until the start of the application experiments.

Results
A total of 18 micro/nanofluidic PDMS devices were batch fabricated with a gap distance and chamber depth of 10.6 6 0.2 and 12.5 6 0.4, respectively (Table 2).The original thickness of the DFR was 15 lm; however, due to thinning of the DFR material during the postexposure bake and development processes, the actual depth of the microchambers was found to be 12.5 lm.Overall, for the 18 micro/nanofluidic devices, the dimensions were controlled within 6% of the design value for the microchambers.For the channels, a total of six devices were fabricated for each desired channel dimension with diameters of 1009 6 33 nm, 523 6 7 nm, and 317 6 6 nm for the microchannel, submicron channel, and nanochannel, respectively (Table 3).The accuracy of the channels' diameters for these fluidic devices was within 6% of the design for the final PDMS devices.It was noted that the dimension of the channel imbedded in PDMS is slightly smaller than the original fiber diameter.This reduction in the channel dimension may be due to the acetone removal process since it has been reported that PDMS swells isotropically in an acetone bath [19].It is anticipated that the isotropic swelling of the PDMS would push against the solvating PMMA fibers, thereby expelling the PMMA out of the channel.Subsequently, when the acetone is removed, the PDMS may not completely return to its original dimension resulting in smaller channel diameters.As a result, the final PDMS channels were determined to decrease by 10.1%, 13.1%, and 1.9% for the 1000 nm, 500 nm, and 300 nm designed channels, respectively.
The top image in Fig. 6 shows two microfibers drawn on the DFR molded microchambers, which demonstrates the ability of this technique to control fiber location on the substrate via the three-axis dispensing system.A detailed characterization study of this direct-writing process has been performed, and an analytical mathematical model has been developed to identify the relevant importance of each parameter in the fabrication process (Table 4).This model will be published in subsequent work.However, our results indicate that fiber length, polymer solution concentration, and dispensing time are significant parameters affecting the final fiber diameter with increased fiber length, lower polymer solution concentration, and reduced dispensing time yielding smaller diameter fibers.
The bottom image in Fig. 6 presents an optical image of a completed PDMS microfluidic platform with two parallel submicron channels, d 1 $ 938 nm and d 2 $ 984 nm, with an interchannel   spacing of $3 lm.An enlarged view of the gap between the two microchambers, referred to as a "dam" since it blocks the flow of fluids and/or particles between the two microchambers, can be seen in Fig. 7.By controlling the diameter of the submicron/nano channel in the dam, the movement of particles between the microchambers can be regulated based on particle size.

Conclusion
In this new fabrication process, we have used PMMA fibers as a sacrificial material in conjunction with dry film photolithography and replica molding to create micro-and/or nanoelectroporation platforms.The fabrication techniques described herein present a new capability for creating micro/nanoscale fluidic platforms with high accuracy.Compared to other traditional fabrication methods, the major advantages of this fabrication process include the ease of fabrication, low cost, high potential for automation, and the use of a nanofiber as a sacrificial structure for creating the nanochannel.In particular, contrasting our technique to soft photolithography, the method presented here simplifies the manufacturing process by: (1) avoiding multilayer photolithography; (2) only needing one photomask instead of multiple photomasks of different dimensions; (3) the fabricated mold can be used repeatedly; and (4) high yield ($95%).The method presented in this paper repeatedly created micron to nanoscale channels imbedded in PDMS within 6% of the channel design values.Having the ability to readily and easily change the mold structures and micro/nano fibers' position and sizes aids in the production of customizable micro/nanofluidic platforms and directly addresses specific requirements and needs for particular biological applications.The platforms developed in this paper are currently undergoing feasibility testing to demonstrate their ability to perform biomolecule transport and nano-electroporation in cells.

Fig. 2
Fig. 2 Flowchart of the fabrication process for the micro/nanofluidic devices

Fig. 3
Fig. 3 Custom-made collimated UV light source and exposure system

Fig. 5
Fig.5Schematic of the direct-write process for creating the micro/nanofiber using three-axis robotic dispensing system

Fig. 6
Fig. 6 Top image: Scanning electron microscope (SEM) image of the dry film resist mold on a glass substrate with PMMA microscale fibers (top fiber: d 1 5 1.96 lm and bottom fiber: d 2 5 1.74 lm).Bottom image: optical microscope image (1003) of a PDMS device with two embedded submicron channels (top channel: d 1 5 938 nm and bottom channel: d 2 5 984 nm) after etching the fibers via sonication in an acetone bath, the separation distance between the two submicron channels is 3 lm.

Fig. 7
Fig. 7 Confocal image of the PDMS device with the dam between the two microchambers and the submicron channel ($954 nm) embedded in the dam

Table 2
Comparison between the design dimensions and actual dimensions achieved for the PDMS microchambers for all the 18 micro/nanofluidic PDMS devices

Table 3
Comparison of fiber diameters to actual channel diameters for a total of 18 devices with six devices per size group