Fabrication and Electroosmotic Flow Analysis of Freely-Suspended, 3D Microchannels

Suspended microchannels suitable for electroosmotic flow were fabricated by coating and dissolving sacrificial polymer fibers created by a new technique which involves using a pressurized syringe loaded with solution to directly deposit filaments of solution on the substrate in a “connect-the-dots” style. Additionally, because both the stylus and syringe were controlled with an ultra-high precision instrument, the fibers, and subsequent microchannels, can be precisely positioned in three dimensions. The diameters of the fibers were controlled, within the range of 400 nm to 100 μm, by varying the concentration of the PMMA solution or the fiber length. It was discovered that increasing either of these variables led to an increase in fiber diameter. The fibers were coated with a hydrophilic layer of glass followed by a structurally-supportive layer of poly(para-xylylene) (Parylene), then dissolved to produce hollow microchannels with diameters ranging from 4 to 100 μm. Electric potentials were applied across buffer solution-filled microchannels suspended between two electrodes to induce electroosmotic flow. A particle imaging velocimetry (PIV) system was employed to visualize and measure flow under potentials ranging from -100 to 100 V. INTRODUCTION Microfluidics-based lab-on-a-chip (LOC) devices and micro total analytical systems (μTAS) are emerging as powerful tools to aid in medical diagnosis, cell sorting, DNA analysis, protein separation, and potential drug evaluation. However, conventional microfabrication techniques have impeded the development of more complex microfluidic systems due to the constraint that fluidic structures must be confined to a single plane. The development of a process to fabricate precisely-positioned, three-dimensional, microfluidic networks would eliminate this constraint and enable advancement in the state-of-the-art of LOC and μTAS devices. BACKGROUND Conventionally, microfluidic channels have been fabricated by selectively etching open troughs in substrate materials, including glass, silicon and metal, and bonding these substrates together to form hermetically sealed conduits [1]. Unfortunately, because these techniques are all based upon traditional planer photolithography, the channel layout is quite limited since individual microchannels cannot cross one another without forming an intersection. Recently, novel techniques have emerged which enable the fabrication of three-dimensional microfluidic networks. Among the most popular of these methods is a process by which several planar polymer monoliths containing microchannels are fabricated via polymer casting onto microfabricated master copies [2-4]. 3D microfluidic structures have also been created by utilizing focused laser pulses to selectively ablate material in a bulk glass substrate [5]. These fabrication techniques have lead to the creation of microand nanofluidic systems and stimulated the development of a variety of novel devices including a selective cell patterning device [6], and a microfluidic computer [7]. Additionally, electrospinning has been employed to produce sacrificial polymer fibers subsequently used as sacrificial structures in the production of nanochannels for the development of a single molecule detector [8]. Although electrospinning is a useful technique, it is not designed for accurate fiber placement between two defined points. Recently, a new family of processes, which involves harnessing surface tension to thin liquid filaments of solvated polymer into solid microand nanoscale fibers, has been recently developed by our group [9,10]. Specifically, Harfenist et al. [9] demonstrated a brush-on fiber drawing technique for polymer fiber “spinning” that allows a user to fabricate microor nanoscale fibers by drawing a polymer solution across a substrate with pre-positioned features, i.e. an array of microfabricated pillars. Filaments of solution adhere to the tops of the pillars, then thin via surface-tension driven necking and dry to yield aligned arrays suspended microand nanofibers. More recently, a partially automated alteration of the manual brush-on technique has been developed and characterized [10]. In this process, a stylus is dipped into a pool of solvated polymer solution, and the solution adheres to the stylus tip. As the stylus is removed, a small filament is drawn between the solution pool and the stylus tip and detached from the stylus by submerging the stylus into a second pool of solution. Because the stylus is controlled by an ultra-high-precision micromanipulator, nanoscale positioning of the fibers in three dimensions was achieved. The focus of this work is to utilize the new techniques developed by our group for fabricating precisely-positioned polymer fibers in the construction of sacrificial structures which serve as scaffolding in the production of 3D microchannels. METHODOLOGY Polymer solutions were made by mixing 996 kDa poly(methyl methacrylate) (PMMA) (Sigma-Aldrich) with chlorobenzene (Sigma-Aldrich) and sonicating for 4 days to ensure complete dissolution. Solutions of 22%, 23%, 24%, 25%, and 26% (all by weight) were produced based on preliminary experimentation that illustrated the capability to produce fibers within this concentration range. Furthermore, 996 kDa PMMA was chosen due to previous success during characterization of the stylus-drawn method [10]. In previous studies [11], microscale fibers were created by using a stylus to draw liquid filaments of solvated polymer from manually-deposited pools. However, diameter variation was found to be quite large due to the buildup of dried solution of the stylus. This caveat led to the development of an alternative direct-write method, in which solution is directly deposited and drawn by a spring-loaded syringe (Figure 1), located ~500 μm above the substrate, that is controlled by a programmable, custom-made, ultra-high-precision micromilling machine (MMM) (Dover Instruments Inc.) with 1.25 nm resolution in the horizontal plane and 20 nm resolution in the vertical direction. Figure 1 Direct-write drawing of PMMA fibers. a) Load syringe with solvated PMMA. b) Pressurize syringe to expel solution from needle into contact with substrate. c) Translate syringe to desired endpoint. d) Pressurize syringe to allow PMMA solution to contact substrate. In order to optimize the direct-write fiber fabrication technique, it was first characterized using the different concentrations of PMMA described above. Fiber diameter and yield, defined as the ratio of unbroken fibers to fiber drawing attempts, was measured under a variety of experimental parameters including concentration, fiber length, and drawing rate. In previous studies using the stylus-drawn method, fiber diameter was characterized for varying concentrations of PMMA and the fiber length was held constant, 2 mm. However, we theorize that fiber diameter can be further influenced by modulating the drawing length. It is expected that viscoelastic stretching will cause a reduction in diameter for longer fibers, perhaps enabling the fabrication of sub-microscale structures. To investigate this hypothesis, the MMM was programmed to draw multiple fibers in series with lengths ranging from 2 mm to 20 mm in 2 mm increments. Additionally, because the solution filaments are rapidly drying during the drawing process, it is expected that drawing rate will also influence the final fiber diameter as well. Therefore, the MMM was programmed to draw fibers at 5 and 20 mm/s. The experiments were repeated four times for each operating condition, i.e. fiber length, stylus speed, and solution concentration. Fiber diameters were measured using an SEM (Zeiss Supra 35VP) in three locations, at the middle of the fiber and 100 μm from each end. In order to enable electroosmotic flows, the channels were fabricated in situ between two gold electrodes, ‘machined’ via conventional photolithographic processing techniques, separated by a micromilled trench (Figure 2). Sacrificial polymer fibers were fabricated using the direct-write method described above. The microchannels were created by sputter coating (RF sputtering) the direct-write fibers with a thin film (20-30 nm) of borosilicate glass (BSG) to establish a hydrophilic inner channel wall. Due to the fragility of the BSG inner wall, a reinforcing outer wall (t=10 μm) of Parylene was deposited using a room temperature, vapor-phase deposition technique (SCS Parylene Deposition System 2010). Following Parylene deposition, access holes were drilled using the MMM and the PMMA was removed to complete the fabrication process. The PMMA removal process consisted of several steps including: 1) dissolution in acetone or chloroform; 2) thermal degradation; and, 3) residual solvent dissolution using critical point drying in liquid CO2 (Tousimis). Figure 2 – Fabrication of channels from PMMA fibers. a) PMMA fiber between two PMMA attachment points on glass substrate with gold electrodes. b) Deposition of BSG and parylene. c) Drilling of 500 μm access holes to expose PMMA. d) Dissolution of PMMA with acetone. To characterize the fabrication process of the microchannels, representative channels were frozen in liquid nitrogen and cleaved to enable VPSEM imaging of the channel interior to confirm complete removal of the polymer. The BSG layer was evaluated following deposition via profilometry (Veeco Dektak) and variable pressure scanning electron microscopy (VPSEM) (Zeiss EVO) under a variety of different conditions including high-power deposition (300 W), low-power deposition (150 W), and multi-layered deposition. EO Flow was induced by loading the microchannels with a phosphate buffer solution (PBS, 10 mM, pH 6.1) and applying a voltage ranging from 0 V to 100V in both polarities between the ends of the channel. This flow was visualized by seeding the buffer solution with 1 μm fluorescent microparticles (Duke Scientific) and imaging the motion of these particles with a particle imaging velocimetry (PIV) system (TSI). The PIV system utilizes a fluorescent microscope (Zeiss) and twin pulse Nd:YAG lasers (Big Sky) to capture two laser-induced fluorescent exposures of the particles in the channel in rapid sequence (100 μs to 100 ms elapsed time between exposures), then correlates the two images to elucidate the velocity profile. RESULTS AND DISCUSSION Fibers with diameters ranging from 403 nm to over 100 μm were drawn using the direct write technique (Figure 3). A two-factor ANOVA test was performed on the 5 and 20 mm/s draw rate data for concentrations ranging from 22% to 26% and lengths ranging from 4 mm to 20 mm. It was determined that an increase in fiber length resulted in a decrease in fiber diameter (P<0.05) and an increase in solution concentration resulted in an increase in fiber diameter (P<0.005) (Figures 4 and 5), which correlates well with previous characterization studies of the stylusdrawn method [10]. Figure 3 – PMMA fibers fabricated with direct-write technique. a) Several fibers drawn in series, b) middle section of single fiber, and c) end of single fiber. Direct Write Characterization 5 mm/s