Spatially and optically tailored 3D printing for highly miniaturized and integrated microfluidics

Abstract

Traditional 3D printing based on Digital Light Processing Stereolithography (DLP-SL) is unnecessarily limiting as applied to microfluidic device fabrication, especially for high-resolution features. This limitation is due primarily to inherent tradeoffs between layer thickness, exposure time, material strength, and optical penetration that can be impossible to satisfy for microfluidic features. We introduce a generalized 3D printing process that significantly expands the accessible spatially distributed optical dose parameter space to enable the fabrication of much higher resolution 3D components without increasing the resolution of the 3D printer. Here we demonstrate component miniaturization in conjunction with a high degree of integration, including 15 μm × 15 μm valves and a 2.2 mm × 1.1 mm 10-stage 2-fold serial diluter. These results illustrate our approach's promise to enable highly functional and compact microfluidic devices for a wide variety of biomolecular applications.

Fig. 1. 3D printed membrane valves and squeeze valves.

a Schematic diagram of membrane valve geometry with cut-away schematics showing a membrane valve in (b) open and (c) closed states depending on pneumatic pressure applied through the control channel. d–o Side- and top-view microscope images of (d–g) 20 pixel, (h–k) 12 pixel, and (l–o) 6 pixel diameter valves in their open and closed states. p Schematic diagram of squeeze valve geometry with cross-section diagrams (rotated 90^∘) showing a squeeze valve in (q) open and (r) closed states depending on pneumatic pressure applied through the control channel, which squeezes together to close the flow channel. s–z Microscope images of (s–v) 4 × 4 pixel (side- and top-view), (w,x) 3 × 3 pixel (top-view), and (y,z) 2 × 2 pixel (top-view) valves in their open and closed states. All scale bars are 100 μm.

Fig. 2. Deliberate exposure time and layer thickness variation in 3D printing process.

a Membrane valve designed dimension parameters, (b–d) membrane valve exposure times. e Squeeze valve designed dimension parameters, (f–h) squeeze valve exposure times. Tables 1 and 2 specify the variable layer thicknesses used for each type of valve.

Fig. 3. 3D printed membrane and squeeze valve pumps.

a, f, k Schematic diagrams. b, g, l Side view microscope photos. c, h, m Bottom view microscope photos. d, i, n Volumetric flow rate as a function of the pump phase interval, Δt. e, j, o Pump volume per cycle as a function of pump phase interval. All graphs were obtained from testing at least three different pumps.

Fig. 4. Fast diffusion mixing.

a Two thin fluid sheets in a narrow vertical channel. b Diffusion time for example molecules over a 15 μm diffusion length in an aqueous solution. D is the diffusion coefficient. Blue = 30 kDa protein, Orange = fluorescein, Green = dissolved gas molecules. c Cross-section scanning electron microscope (SEM) image of narrow 3D printed diffusion mixing channel.

Fig. 5. Single stage 1:1 mixer.

a–e Membrane valve-based pump version: schematic diagram (a) perspective view and (b) top view, (c) microscope photo, (d) mixing test, and (e) time to equilibrium test. Note in (d) that the mean, ${\overline{C}}_{rel}$, and standard deviation, σ, of the relative concentration are both plotted on the left vertical axis. f–h Squeeze valve-based pump version: schematic diagram (f) perspective view and (g) top view, (h) microscope photo.

Fig. 6. 10-stage twofold serial diluters.

a CAD drawing and (b) microscope image of membrane valve-based serial diluter. c Normalized fluorescein concentration as a function of time for all ten output channels (outputs 1–10) and the input concentration (output 0). d Steady state normalized fluorescein concentration at each output channel for three repeated tests. e Microscope image of 10-stage twofold serial diluter made with squeeze valve-based 1:1 mixer modules.

Fig. 7. Digitonin permeabilization assay.

a μCT image of 3D printed 5-stage diluter integrated with 3D printed cell plate. b Fluorescence images of serially diluted fluorescein. The experiment was repeated three times and all show similar results. c Cell treatment workflow. d Whole well images of differentially treated A549 cells with propidium iodide as a marker (red). The treatment fluid was 100 μg/mL digitonin and 2 μM propidium iodide in DMEM/F12 while the control fluid was 2 μM propidium iodide in DMEM/F12. 100% ethanol was used as a positive control for the whole experiment. e The semi-log dose response-curve was derived from the experiment. Responses were quantified through the measurement of propidium iodide area relative to the total cell area (DIC). Bounds were set at 0 μg/mL digitonin (0 response) and 100% ethanol (100% response). Values were derived from n = 3 independent experiments. Error bars denote standard deviation.