P44: Toward 3D-Printed Microfluidic Artificial Lungs with Complex, Biomimetic Geometries

Abstract

Purpose: Current hollow-fiber oxygenators used in Extracorporeal Membrane Oxygenation have suboptimal gas exchange performance and hemocompatibility. Microfluidic artificial lungs (µALs) promise smaller feature sizes, improving gas exchange efficiency, and biomimetic flow paths, improving blood compatibility. For the first time, we leverage the geometric freedom of 3D-printing technology to demonstrate µALs with truly 3D, biomimetic branching capillaries and gas-side designs to increase gas exchange surface area. Methods: Two µAL designs were generated in Solidworks and printed on an Asiga MAX X27 UV printer. The first was a “simple branching” µAL (Fig 1A) with quadfurcating, biomimetic branching for blood distribution. The second was a “hatched” µAL (Fig 1B) employing a gas phase with strategically placed supports to maximize gas exchange surface area while maintaining structural stability. These changes are only to the gas, and so the blood-phase is unchanged. Dyed water was used to visualize internal structures (Fig 1D). Geometric analysis and proven models were used to calculate theoretical shear rate and rated flow. Computational Fluid Dynamics (CFD) was used to visualize flow (Fig 1E,F). Experimental pressure drop vs water flow rates were measured (Fig 2). Results: Through geometric analysis (Table 1), branching distribution technique provides more uniform shear stress (0.95-1.34 dyn/cm2) compared to a non-branching “open-plenum” control (0.09-1.27 dyn/cm2). The hatched µAL has a 3.18 mL/min rated flow, 1.7x greater than the 1.84 mL/min rated flow of the non-hatched branching µAL. As a result of the higher rated flow, it also has higher theoretical shear of 1.64-2.31 dyn/cm2 (at rated flow) despite identical blood-side geometry. The branching lung demonstrated a pressure drop up to 12 mmHg at flows up to 5 mL/min (Fig 2). These are significantly larger than analytically predicted values (extrapolated 7 mmHg vs 0.233 mmHg at 3.2 mL/min), prompting investigation into resolution of internal capillaries and effects of surface roughness on pressure drop. Conclusion: We have demonstrated the first 3D-printed branching biomimetic µAL formed from a gas permeable material, with gas-side geometries to improve gas exchange efficiency. These geometries are expected to translate to greater blood flows enabling creation of future animal and human-scale µALs. Table 1. Theoretical calculations. All uALs have 256 8.5mm long 216x324 µm blood and gas channels, with 54 µm-thick membranes. They differ in their effective gas exchange area, and blood distribution method, resulting in differing rated flows, shears, and pressure drops. The Branching uAL employs a branching distribution, resulting in a much tighter shear stress range. The GasHatch uAL employs both a branching distribution, and a blood-gas geometric-synchronization, with 100% rather than 58% of the gas-exchange capillary face being used for gas exchange. Values for a theoretical “control” lung (not fabricated; contains an open plenum/ non-branching distribution region) are displayed for comparison. Flow velocity and shear stresses are calculated separately for capillary and distribution areas.Figure 1. Blood flows are indicated using red arrows, and gas flows are indicated using blue arrows. (A) Design of simple (non-hatched) branched μAL. (B) Close-up of the hatched μAL capillaries and gas channels, showing how gas-side supports are only in regions not adjacent to capillaries and therefore not reducing gas exchange surface area. (C) Hatched μAL with the gas phase filled with blue dye for visualization. (D) Non-hatched branching μAL with blood phase filled with blue dye, and gas phase filled with red dye. (E) CFD visualization of flow velocities in the branching blood distribution network; (F) CFD analysis of pressure drop in the biomimetic blood distribution channels.Figure 2. Pressure drop of branching µALs (water; n=3)