Expanding the droplet microfluidic community – towards a modular active platform

Abstract

Microfluidics has progressed tremendously as a field over the last two decades. Various areas of microfluidics developed in fully-fledged domains of their own such as organ-on-a-chip, digital and paper microfluidics. Nevertheless, the technological advancement of microfluidics as a field has not yet reached end-users such as chemists and biologists for independent use. A modular automated platform is envisioned to provide the stacking and modularity required to lower the knowledge barrier for end-users; hence, microfluidics will simply be considered as a tool for the application-focused researchers. The numerous advantages of droplet microfluidics–self-contained reactions, shorter reaction times, lower reagent consumption–will be leveraged by non-microfluidic researchers. However, the technological and knowledge gaps between the current state of the field and the modular automated platform is prohibitively large. This thesis aims to significantly advance the automated technology and to target key issues restricting droplet microfluidic accessibility. The main advancements are separated into two categories: technological and knowledge-focused. The technology-focused advancements (semi-automated droplet control, open-source pressure pump, critical system overview) are necessary stepping stones towards the development of a fully automated modular microfluidic platform. The knowledge-focused contribution (air tubing dynamics, droplet resistance, microfluidic chip compliance) enable the smarter development of the technology; better understanding and modelling the system is of special importance for active control. Furthermore, a deeper understanding of the system can be leveraged for other active platforms and passive microfluidic devices.

The semi-automated droplet manipulations are implemented in an additional layer of the control algorithm. The functionality enables a user to set the droplet length or split ratio before automatically performing the manipulation. The droplet generation accuracy is ± 10 % of the length and a monodispersity of ± 1.3 % for 500-µm-long droplets. The splitting ratio resolution is limited by the channel width for the daughter droplet length; the accuracy is ± 4 % of the initial length. The droplets are merged on-demand. Finally, the effective mixing of the droplet is demonstrated. The manipulations are leveraged in a qualitative drug screening assay that showcases the potential of the platform.

µPump is an open-source pressure pump that targets microfluidic users. Researchers focusing on either passive or active microfluidics can benefit from this system. A similar application performance (i.e. consistency in droplet volume) is achieved for a lesser price tag than comparable commercial systems. Furthermore, the open-source nature of the system enables a better understanding of the actuation limitation and the communication protocol.

The air tubing connecting the pressure pump outlet to the reservoir holder is generally neglected. The dynamics of the tubing are investigated. For 1/16” inner diameter, the dynamic effects are negligible for a length up to 60 cm and a pressure resolution of 2 mbar. For the 1 mm inner diameter tubing, the dynamic effects are significant. The dynamics are modelled as a first-order system. The performance difference between the nonlinear and linear first-order model is found to be negligible. Therefore, a simple first-order model with a time constant depending on tubing length is deemed adequate. Passive and active microfluidic devices benefit from a better understanding of the air tubing dynamics.

Droplet resistance affects the micro-channel network behaviour as they move through the channels. The uncertainties of their impact lengthen the iterative design process. Numerous studies investigated droplet resistance. However, a consensus is still pending. Most methods rely on passive principles. Oppositely, the method herein introduced relies on the active droplet control platform and grey-box system identification. The preliminary results are promising and in agreement with the literature. Process improvement is envisioned to better the resolution and to enable apply the technique to many more conditions such as non-Newtonian fluids.

The model of the microfluidic chip compliance is improved and justified using experimental results. The better understanding of the dynamics is especially impactful for active microfluidics but also for passive microfluidics. A fitting parameter (φ) is required to adjust for the difference in geometry and viscosity. The capacitance is on the order of 10−15. The previous formula predicted values around 10−20 whereas the other formula (without considering the fitting parameter) predicts around 10−16. Moreover, the relationship of the fitting parameter is unintuitively stronger with respect to the height-to-length ratio than with the width-to-length ratio. The length is related to the flow rate. Shorter lengths mean larger flow rates, and consequently, larger volume per unit pressure (i.e. capacitance).

The path towards an automated modular platform that can easily be adopted by end-users as a tool relies on a shift towards a standalone system. The most important components to focus on are the actuation, feedback system, and the automation algorithm. The actuation through the current pressure pump is limiting due to its dependence on a pressured airline. The bulky and expensive microscope prohibits users from easily adopting the platform. Finally, the algorithm must be further developed to handle the procedure automatically such that minimal user input is required.