Microchip-based capillary electrophoresis could yield faster analysis times with lower reagent consumption, easier multiplexing, and greater ease of use than CE in silica capillaries. However, the glass microchips commonly used are expensive to manufacture, requiring extensive fabrication facilities, and can be ill-suited to applications for which cross-contamination is an issue and single-use devices are desired. In contrast, plastic, or polymeric microfluidic chips can be manufactured with imprinting or molding techniques with relatively minimal equipment, and manufactured with hot-embossing or injection molding techniques for pennies per chip. However, laser-induced fluorescence detection in polymeric microchips presents some unique challenges. Because the plastic substrate is substantially more fluorescent than freestanding silica capillaries, spatially selective detection is required to isolate the fluorescent signal originating from within the channel in order to achieve the desired sensitivity. In the past, this has required a confocal system, with the measurement of multiple channels achieved by mechanical scanning of the optical elements. We have previously developed and demonstrated a new scheme for sensitive, spatially selective and spectrally resolved laser-induced fluorescence detection from multiple microfluidic channels, and applied this scheme to 10 Hz five-color forensic DNA analysis in a polymeric microfluidic device. Free-space 488 nm laser excitation is spread into a collimated line via two cylindrical lenses and then split into multiple focused spots using an array of spherical plano-convex lenses with diameters equal to the microchannel spacing. At each excitation spot, a ball lens and optical fiber combination is positioned underneath the microchannel. The spatial selectivity is achieved by using a high refractive index ball lens and a substantially smaller-diameter optical fiber positioned to obtain focused light from the channel. The detection optics can be freely positioned near each channel, placing minimal constraints on channel layout and design. The other ends of the optical fibers are formed into a 1-D array and directed onto the entrance slit of an imaging spectrograph. Analysis of standard DNA base-pair ladders in an eight-channel configuration shows comparable sensitivity to that obtained with measurements of a single channel using a commercial confocal microscope. The limit of detection is approximately 10pM for fluorescein in a single polymeric channel. The prototype instrument is robust, versatile, contains only fixed optical parts, and has the potential to be more cheaply implemented than competing technologies. The economies of parallel detection and the importance of spatial selectivity make this method generally useful for separations in polymeric substrates with multiple microchannels. Although this technology has been evaluated using short-tandem repeat DNA separations, the instrument can easily be used for most multi-color, multi-channel CE analyses. We have assembled a duplicate instrument in order to address this problem, and transferred the technology for fabricating the polymer microchips to our facilities at NIH. We have also worked to optimize recipes for patterning and bonding microchips in different polymer substrates, such as clinical quality PMMA, polycarbonate, and PDMS for our equipment. Last year, we started using a UV-ozone activation step prior to device bonding in lieu of the solvent assisted process used previously. This adjustment led to substantially higher device yield as well as greater reproducibility in channel cross section. This year, we started using a bonding method that employs a sacrificial material to protect the channels during the bonding process, a change that we hope will further improve device yield and bond strength, enabling the use of more aggressive chemistries for treating the channel walls to minimize non-specific interactions. Currently, we are treating the walls of the channels with methyl cellulose, which substantially reduces interactions between the labeled peptides and the channel walls, but there is some degradation of the coating over time. This year, we have also continued to optimize buffer conditions for the separations. The most important additional constraint for separations in plastic microchips, as opposed to glass ones, is to keep the overall channel conductivity low, as the lower thermal conductivity of the substrate can give rise to peak broadening from Joule heating at substantially lower dissipated power densities than in glass devices. As a result of our work, the analyte peak widths were reduced by up to a factor of forty, and are now within a factor of three of the limit given by diffusional broadening of the injection plug. Using our laboratory-built setup, we have successfully separated nanogram-level quantities of several fluorescently labeled neuropeptides in less than two minutes. In addition, we have begun experiments aimed at implementing an on-chip immunocapture step prior to electrophoretic separation. The results of preliminary experiments using fluorescent microscopy to verify the success of the surface chemistry, and separations using similar buffer conditions as needed for release of the capture neuropeptides, are promising.
