Computational toolset for fluidic network-to-layout translation applied in FCB design for chemical gradients in parallelized organ-on-chips

Microfluidic devices improve automation and integration of complex biological and chemical workflows. Integration often requires connection of functional units via networks of microfluidic channels in an overarching manifold, the design of which typically demands substantial domain expertise and multiple iterations. The network-of-components topology resembles architectures used in printed circuit board design. Although hydraulic networks and electronic circuits are analogous [1], fluidic channels have substantial hydraulic resistances compared to near-ideal conductors, which adds design complexity. These hydraulic resistances and their footprints must be considered when solving a network to satisfy flow and pressure constraints for component functioning. Accounting for this, concepts from electronic design automation can be leveraged to develop a novel method to design microfluidic channel networks.
Here, we present an open-source software tool that, given a target network, fabrication constraints, and component specifications, automatically produces a manufacturable microfluidic device layout. Components are automatically placed and connected with fully dimensioned channels. This results in a straightforward design process for intricate microfluidic channel networks required in advanced biological and chemical experiments. We applied the tool to design a parallel, combinatorial Organ-On-Chip (OoC) platform that enables soluble-compound testing over a broad range of concentrations and mixtures on multiple cell-culture chips with membrane-separated compartments (Fig.1), while ensuring transmembrane pressure (TMP) matching. These chips are commonly used to model tissues such as lung, gut, brain, and joints [2, 3]. To create a log10-scale compound concentration gradients, we integrated a passive, mixer-based serial diluter. As the primary network component, we designed and CNC-micromachined a herringbone micromixer (Fig.2A-C) and validated its mixing efficiency using simulations and fluorescence imaging (Fig.2D). Next, we formulated a multilayer fluidic circuit board (FCB) network description in which micromixers and chips are interconnected in a serial-dilution architecture. We specified connection positions, desired flowrates for the chips and mixers to achieve the required concentrations, and layer assignments for the various fluidic paths (Fig.3A). Using the component properties and network description, the software applied iterative modified nodal analysis [1] to calculate the pressure drop over each channel, and selected channel dimensions that result in the required hydraulic resistances while considering fabrication limits on feature size and aspect ratio (Fig.3B). The resulting layout fits within a standard well plate footprint. The tool exports per-layer DXF-files used to create a CAD design for CNC milled parts (Fig.3C). After milling, the FCB was assembled, the chips were clamped onto the ports, and tubing was attached to the in- and outlets. Syringe-pump experiments with food-dye solutions showed that the tool-computed inlet flow rates produced a wide concentration gradient across the three chips (Fig.3D). To conclude, integrating the proposed computational toolset allows us to design complex fluidic circuits far more efficiently. This acceleration of the design cycle significantly expands the range of problems that can be practically solved, and lowers the barrier to entry for non-domain users interested in microfluidics. In combination with open platforms such as TOP, which integrate the peripherals, including control electronics, pumps, and reservoirs, our approach can increase the accessibility of microfluidics [4, 5].