Over the past decade, various microfluidic fluorescence-activated cell sorter (FACS) systems have been demonstrated aiming to provide a fully enclosed environment for sterile, contamination and infectious free sorting, and better downstream microfluidic integration for further analysis after sorting. The biggest challenge of μFACS systems, however, is the orders of magnitude lower sorting throughput and purity than commercial aerosol based FACS (90% purity at 70,000 cells/sec). To solve this problem, the key innovation is to find a faster switching mechanism. In this dissertation, the importance of the cell sorting in academic research and clinical practice will be discussed at first. Among the commercial FACS machines, the state-of-the-art performance is achieving above 90% purity at 70 000 cells/sec. Other reported μFACS presents 1~2 order of magnitude lower in throughput.
In this context, we attempt to use the phenomena of the pulsed laser induced cavitation bubble to switch particles/cells in high speed. The first version simply excited the bubble in the microfluidic channel that is running parallel to the sample channel. The bubble expansion deformed the elastic PDMS channel wall and induced fluid swing in the sample channel which carried target particle/cell flowing into the collection channel. This first trial demonstrated the fast dynamics of this switching mechanism and the strong power the cavitation bubble can affect the flow. However, the deformation of the channel wall extended over hundreds of microns in length, resulting in low post-sort purity. To overcome this, the second version designed a connecting nozzle between the sample channel and the bubble channel. During the expansion of bubble, a liquid jet was created through the nozzle to the sample channel, which limited the disturbance range to tens of microns. With 2D hydrodynamic focusing, the pulsed laser activated cell sorter (PLACS) demonstrated performance with above 90% post-sort purity at 3 000 particles/sec. The drop of post-sort purity at higher throughput was due to the lack of the third dimension focusing which created large time variation in cell arrival time between the detection and switching zones. Then 3D hydrodynamic focusing that utilized multilayer PDMS with vertical vias solved the synchronization issue but also allows efficient particle switching using a smaller bubble with a smaller perturbation volume, and a shorter on-off switching cycle. As a result, the 3D PLACS was able to sort at 23 000 cells/sec with ~ 90% purity or at 45,000 cells/sec with 45% purity within a single channel. However, the large amount of diluting sheath fluid needed for tight 3D focusing requires a high initial cell concentration in samples (>107 cells/ml), which can result in cell clogging in channels, and high pressure pumps to drive fluid through microchannels at high speeds. Then inertial focusing was employed as a substitute to hydrodynamic focusing. The integrated system achieved sorting at 10 000 particles/sec with >90% sort purity or 6 000 cells/sec with >80% sort purity and used 10 times lower initial concentration cell samples than that in sheath-based PLACS.
The achieved PLACS not only outperformed other μFACS but also achieved performance at a level comparable to conventional aerosol-based FACS. It is anticipated that by integrating with upstream and downstream microfluidic cell analysis functions, the developed PLACS will greatly facilitate biomedical research and clinical diagnostics.