Miniaturization, Parallelization, and Automation of Endotoxin Detection by Centrifugal Microfluidics.

We demonstrate microfluidic automation and parallelization of Limulus amebocyte lysate (LAL)-based bacterial endotoxin testing using centrifugal microfluidics. LAL is the standard reagent to test for endotoxin contaminations in injectable pharmaceuticals. The main features of the introduced system are more than 90% reduction of LAL consumption, from 100 µL/reaction to 9.6 µL/reaction, automated liquid handling to reduce opportunities for contamination and manual handling errors, and microfluidic parallelization by integrating 104 reactions into a single centrifugal microplate. In a single Eclipse microplate, 21 samples and their positive product controls are tested in duplicate. In addition, a standard curve with up to five points is generated, resulting in a total of 104 reactions. Test samples with a defined concentration of 0.5 endotoxin units per milliliter were tested, resulting in a coefficient of variation below 0.75%. A key feature for achieving a small coefficient of variation is ensuring the same path length along the microfluidic channels to the final reaction chambers for each sample and the reagent, so that any unspecific adsorption to the polymer surfaces does not affect the accuracy and precision. Analysis of a sample containing naturally occurring endotoxin with the developed microfluidic microplate yielded comparable results to the conventional testing method. A test with eight commercially available pharmaceuticals was found to pass all requirements for bacterial endotoxin testing as specified in the United States Pharmacopeia. The automated endotoxin testing system reveals specific advantages of centrifugal microfluidics for analytical biochemistry applications. Small liquid volumes are handled (metered, mixed, and aliquoted) in a very precise, highly integrated, and highly parallel manner within mass-fabricated microplates.

Monte Carlo computer simulations and electron microscopy of colloidal cluster formation via emulsion droplet evaporation.

We consider a theoretical model for a binary mixture of colloidal particles and spherical emulsion droplets. The hard sphere colloids interact via additional short-ranged attraction and long-ranged repulsion. The droplet-colloid interaction is an attractive well at the droplet surface, which induces the Pickering effect. The droplet-droplet interaction is a hard-core interaction. The droplets shrink in time, which models the evaporation of the dispersed (oil) phase, and we use Monte Carlo simulations for the dynamics. In the experiments, polystyrene particles were assembled using toluene droplets as templates. The arrangement of the particles on the surface of the droplets was analyzed with cryogenic field emission scanning electron microscopy. Before evaporation of the oil, the particle distribution on the droplet surface was found to be disordered in experiments, and the simulations reproduce this effect. After complete evaporation, ordered colloidal clusters are formed that are stable against thermal fluctuations. Both in the simulations and with field emission scanning electron microscopy, we find stable packings that range from doublets, triplets, and tetrahedra to complex polyhedra of colloids. The simulated cluster structures and size distribution agree well with the experimental results. We also simulate hierarchical assembly in a mixture of tetrahedral clusters and droplets, and find supercluster structures with morphologies that are more complex than those of clusters of single particles.