Continuous-flow microfluidic large-scale integration (mLSI) shows increasing importance in biological/chemical fields, thanks to its advantages in miniaturization and high throughput. Current mLSI is designed manually, which is time-consuming and error-prone. In recent years, design automation research for mLSI has evolved rapidly, aiming to replace manual labor by computers. However, previous design automation approaches used to design each microfluidic layer separately and over-simplify the layer interactions to various degrees, which resulted in a gap between realistic requirements and automatically generated designs. In this paper, we propose a module model library to accurately model microfluidic components involving layer interactions; and we propose a co-layout synthesis tool, Columba, which generates AutoCAD-compatible designs that fulfill all designs rules and can be directly used for mask fabrication. Columba takes plain-text netlist descriptions as inputs, and performs simultaneous placement and routing for multiple layers while ensuring the planarity of each layer. We validate Columba by fabricating two of its output designs. Columba is the first design automation tool that can seamlessly synchronize with the manufacturing flow.

In flow-based biochips, peristaltic pumps consisting of valves are essential to generate circulation flow in a mixer. When a peristaltic pump is activated, the related valves for peristalsis are required to be actuated for many times. However, the roles of valves in traditional chips are fixed, and therefore the valves for peristalsis can wear out much faster than the valves for guiding fluid transportation. This could lead to a reduced lifetime of the chip, because the whole chip function can be affected when just a few or even only a single valve wears out. In this paper, we propose a valve-centered architecture with virtual valves, based on which we introduce a valve-role-changing concept to balance the valve actuations. By switching a valve into different roles, microfluidic components such as mixers, storages, and flow channels can be formed dynamically during the assay process, which enables us to balance the utilization of valves, and synthesize designs that support various kinds of operations. Compared with our preliminary work, we further decrease the largest number of valve actuation as well as the number of valves by the revised dynamic device mapping and fluid path routing. For dynamic device mapping, we introduce a virtual-boundary concept to generate devices at better places while connections between devices are still guaranteed. For fluid path routing, we accurately model valve actuation resulting from our valve-actuation-aware routing, and revise the results by rip-up and reroute. In addition to performance, we improve the reliability of our method by assuring fluid paths from devices to chip boundaries. Experiments show that the new method can be eight times better than the traditional method, and outperforms our preliminary work for large cases even with fewer valves.

This article presents a new concept for designing on-chip storage and cache that will enable efficient management of intermediate fluid samples during transportation and for reuse. By minimizing channel conflicts and recognizing maximum independent sets, storage requirements are efficiently handled jointly by channels as well as by both distributed and dedicated storage cells.

On flow-based biochips, valves that are used to form peristaltic pumps wear out much earlier than valves for transportation since the former are actuated more often, which leads to a reduced lifetime of the chip. In this paper, we introduce a valve-role-changing concept to avoid always using the same valves for peristalsis. Based on this, we generate dynamic devices from a valve-centered architecture to distribute the valve actuation activities evenly and reduce the largest number of valve actuations with even fewer valves. In addition, we propose in situ on-chip storages, which can overlap with other devices, so that less area is needed compared with dedicated storages on traditional chips. Moreover, our method provides good support for assays requiring different volumes and ratios of samples. Experiments show that compared with traditional designs, the largest number of valve actuations can be reduced by 72.97% averagely, while the number of valves is reduced by 10.62%.

Continuous-flow microfluidics have evolved rapidly in the last decades, due to their advantages in effective and accurate control. However, complex control results in complicated valve actuations. As a result, sophisticated interactions between control and flow layers substantially raise the design difficulty. Previous work on design automation for microfluidics neglects the interactions between the control and flow layers and designs each layer separately, which leads to unrealistic designs. We propose the first planarity-guaranteed architectural model, and the first physical-design module models for important microfluidic components, which have modelled the interactions between both control and flow layers, while reducing the design difficulty. Based on the above, we propose the co-layout synthesis tool called Columba, which considers the pressure sharing among different valves, and routes channels in an any-angled manner. Experimental results show that complicated designs considering layer interactions can be synthesized for the first time.

Microfluidics and 3D printing offer exciting opportunities for the development of new technologies and applications in the fields of biology, chemistry, and medicine. However, the design of 3D-printed microfluidic devices remains a challenging and complex task, requiring specialized knowledge and expertise in fluid mechanics, 3D modeling, and 3D printing technology. Currently, there are very few tools helping engineers to do the labor-intensive process of designing microfluidic devices, let alone any tools that can help them design microfluidic devices for 3D printing. In this work, we introduce Flui3d, an interactive software platform for designing microfluidic devices for 3D printing. Flui3d offers a standard parameterized component library, support for multi-layer design, and the ability to design and configure microfluidic devices without the need for specialized knowledge. Flui3d incorporates a distinctive Design-for-Manufacturing (DFM) function, facilitating seamless fabrication of the designed microfluidic devices using commercial consumer-grade printers. We discuss the key features and benefits of Flui3d and demonstrate them by designing examples of microfluidic devices. We also discuss the design complexity and the potential applications of Flui3d.