We describe a highly efficient microfluidic fluorescence-activated droplet sorter (FADS) combining many of the advantages of microtitre-plate screening and traditional fluorescence-activated cell sorting (FACS). Single cells are compartmentalized in emulsion droplets, which can be sorted using dielectrophoresis in a fluorescence-activated manner (as in FACS) at rates up to 2000 droplets s(-1). To validate the system, mixtures of E. coli cells, expressing either the reporter enzyme beta-galactosidase or an inactive variant, were compartmentalized with a fluorogenic substrate and sorted at rates of similar to 300 droplets s(-1). The false positive error rate of the sorter at this throughput was < 1 in 10(4) droplets. Analysis of the sorted cells revealed that the primary limit to enrichment was the co-encapsulation of E. coli cells, not sorting errors: a theoretical model based on the Poisson distribution accurately predicted the observed enrichment values using the starting cell density (cells per droplet) and the ratio of active to inactive cells. When the cells were encapsulated at low density (similar to 1 cell for every 50 droplets), sorting was very efficient and all of the recovered cells were the active strain. In addition, single active droplets were sorted and cells were successfully recovered.

Multiple emulsions are formed using poly(dimethylsiloxane) microfluidic devices. The single emulsions (see image, left) are formed using a single drop maker with uniform wettability. The double, triple, quadruple, and quintuple emulsions (right) are formed using linear sequences of drop makers with alternating wettability.

Microfluidic devices can produce highly monodisperse drops at kilohertz rates using flow-focus drop formation. We use single-layer membrane valves to control, in real time, the dimensions of the flow-focus drop makers. This allows drop size and frequency to be controlled in real time and without adjusting flow rates.

Loading drops with discrete objects, such as particles and cells, is often necessary when performing chemical and biological assays in microfluidic devices. However, random loading techniques are inefficient, yielding a majority of empty and unusable drops. We use deformable particles that are close packed to insert a controllable number of particles into every drop. This provides a simple, flexible means of efficiently encapsulating a controllable number of particles per drop.

We study the impact of inlet channel geometry on microfluidic drop formation. We show that drop makers with T-junction style inlets form monodisperse emulsions at low and moderate capillary numbers and those with Flow-Focus style inlets do so at moderate and high capillary numbers. At low and moderate capillary number, drop formation is dominated by interfacial forces and mediated by the confinement of the microchannels; drop size as a function of flow-rate ratio follows a simple functional form based on a blocking-squeezing mechanism. We summarize the stability of the drop makers with different inlet channel geometry in the form of a phase diagram as a function of capillary number and flow-rate ratio.

We implement image correlation, a fundamental component of many real-time imaging and tracking systems, on a graphics processing unit (GPU) using NVI-DIA's CUDA platform. We use our code to analyze images of liquid-gas phase separation in a model colloid-polymer system, photographed in the absence of gravity aboard the International Space Station (ISS). Our GPU code is 4,000 times faster than simple MATLAB code performing the same calculation on a central processing unit (CPU), 130 times faster than simple C code, and 30 times faster than optimized C++ code using single-instruction, multiple-data (SIMD) extensions. The speed increases from these parallel algorithms enable us to analyze images downlinked from the ISS in a rapid fashion and send feedback to astronauts on orbit while the experiments are still being run.