Microfluidic Cell and Particle Sorting using Deterministic Lateral Displacement

Abstract: Microfluidics is a field which has the potential to revolutionize medical diagnostics. The ability to sort and analyse cells by the use of portable handheld devices has the potential to enable sensitive devices for point-of-care diagnostics. This could play a major role in diagnosis of a wide range of diseases which are prevalent in remote areas in developing countries where a majority of the people live without access to modern health care. One focus of this thesis is the development of such a tool for diagnosis of human African trypanosomiasis (HAT), a deadly disease which requires accurate and rapid diagnosis in order to start treatment at an early stage for better chances of survival. The way the disease is normally diagnosed is by microscopic examination of blood smears in order to see if the blood parasite which causes HAT is present in the patient or not. This is a tremendously difficult task as in a single drop of blood there can be 500 million red blood cells (RBC) while at the same time the number of parasites can be less than 10. The vast majority of background cells makes finding the parasite as difficult as finding a needle in a haystack. To help in diagnosis we present a way by which we separate out the blood parasites, if present, from the blood so they can be easily detected. The method which we use to achieve sorting is called deterministic lateral displacement (DLD), a continuous and passive particle sorting technique which relies on flowing the sample through an array of obstacles. The separation of particles relies on the positioning of these obstacles such that small particles are not affected by the obstacle array while large particles are, and travel at a different angle. The cut-off size between small and large particles is known as the critical diameter and is relatively well characterized for spherical particles. However, the sorting of irregular shaped particles such as RBCs and the long slender parasites has been proven difficult in DLD device. In order to achieve sorting we present a method by which we can control the orientation of particles in order to accentuate their morphological differences. Using this method we present a tool with which we, with high accuracy are able to achieve sorting between these otherwise indistinguishable particles; a tool which could find use in the field-diagnosis of HAT (Paper I and II).We build on the finding of how non-spherical particles, RBCs, behave in DLD devices and combine that with an investigation of how forces exerted on deformable particle act to deform them. This results in the ability to sort based on the size, shape and deformability of a particle, and the successful sorting of normal RBCs and chemically induced, pathologically relevant, forms of RBCs (Paper III). Further on, RBCs are known to undergo cyclic dynamic events such as tumbling and tank-treading. We address this and show that the sorting in DLD is highly dependent on the dynamic behaviour of RBCs which opens up for new sorting schemes based on for example the internal viscosity of RBCs, a factor which is known to influence the choice of dynamic mode (Paper IV). Microfluidic particle sorting has, due to small length scales, often low throughput. This is addressed by investigating the effects of sample concentration on sorting efficiency (Paper V). By changing the sorting behaviour of particles depending on their vertical position in the device, we implement density-based sorting in DLD (Paper VI).DLD can be used to separate any particle mixtures, not only naturally occurring biological particle systems. In Paper VII, we expand the scope to also include thermoresponsive microgel clusters generated by droplet microfluidics which allows for the encapsulation at ~10 kHz frequencies. By allowing the aqueous content of the droplet to evaporate we force the microgels into close contact and can subsequently crosslink them. With the microgel particles which makes up these clusters having temperature dependent interactions we have a way to selectively tune the interaction sites on the clusters. In combination with DLD we can sort out clusters of interest consisting of a specific number of microgel particles which opens up for new types of particles with interesting properties which is highly suitable for interaction studies and the development of novel materials. Microfluidics draws benefit from differences in scaling between variables when length scales decreases from macroscale to microscale, such as the increased surface area-to-volume ratio. This acts to introduce new phenomena which can be exploited as in the examples above. Additional benefits to microfluidics include the ability to handle smaller volumes, faster energy dissipation, precise spatial and temporal control and the ability to integrate multiple preparative and analytical functionalities on a single chip, a so-called lab-on-a-chip, which is a major driving force for the microfluidic field and definitely the direction that I have endeavoured to push the projects within this thesis. For example, the continued development of the parasite sorting device has the potential to not only separate RBCs and parasites but also detect the result, and consequently enable a complete handheld sample-to-answer device dedicated for field-use.