Abstract
Recent advancements in microfluidic biochips allow for easier and faster design and fabrication of increasingly complex biochips to replace conventional laboratories. A roadblock in the deployment of biochips however is their low reliability. Physical defects can be introduced during the fabrication process, and may lead to failure of the biochemical application. This can be costly because of the reduced manufacturing yield, the need to redo lengthy experiments, using expensive reagents, and can be safety-critical, e.g., in case of a cancer misdiagnosis. Researchers have started to propose fault models and test techniques for continuous flow biochips. Six typical defects: Block, leak, misalignment, faulty pumps, degradation of valves and dimensional errors have been identified. The resulting faults can be abstracted into blocks and leaks for simplicity. Both fault types can occur in the control-as well as the flow channel, some common causes being environmental particles, imperfections in molds or bubbles in the PDMS gel. While some faults may be detected before the execution of an application by introducing a test run, other faults occur only during runtime as a result of deterioration or caused by the applied pressure. If such a fault is detected during runtime, e.g. with a CCD camera, we propose a just in time solution that calculates and assigns fluid volumes to alternate components and routes allowing for the completion of the application despite the occurring fault.