<p>Lab-on-a-chip technologies have created a burgeoning number of new and novel devices designed to automate biological processes on-chip in an efficient and inexpensive format for far reaching point-of-care (POC) medicine and diagnostic treatments and for remote and on-line monitoring functions. This work designed a device that integrated advanced optical functionality on-chip with the microfluidics to relieve the reliance on traditional bulky and expensive free-space optics and a high-quality light source. The multimodal input beam was reshaped into an optimized geometry in the microchannel via a 2D system of lenses - improving the quality and reliability of detection through uniform detection of particles. A uniform beam geometry across the sample stream with a uniform beam width will allow repeatable excitation and burst duration to allow for more reliable and predictable detection. Numerous beam geometries were created and the quality and illumination properties confirmed by testing each with a couple sizes of fluorescent and non-fluorescent microspheres to test the effect of beam geometry and particle size combination on device performance. The measured coefficient of variation (CV) for fluorescent beads was found to have a particular beam geometry that yielded best device performance based on the bead size. Fluorescent beads 2.5µm in diameter had a CV of 8.5% for a 3.6 µm beam waist while 6 µm beads yielded a 14.6% CV with a 10 µm beam waist. When measuring scatter and fluorescence signal from a 10 µm the 2.5- and 6.0 µm beads gave 11.4% and 15.8% and 15.9% and 20.4% fluorescent and scatter CVs for each set of beads, respectively. Separately testing each beam geometry with 1-, 2-, and 5 µm beads did not yield any predictable ideal beam-bead ideal pairing for best performance. Lastly, further integration of optical function was shown through the on-chip collection of signals; CVs of 29% and 30% were measured for side scatter and forward scatter, respectively, for 5 µm beads. The reliability of this all-optically guided scheme was confirmed by comparing it to a simultaneously recorded free-space collection scheme. The coincidence rate was found to be 94% and 96% for the side scatter and forward scatter schemes. Both had very low false positive rates – below 0.5% - with missed detection rates that were satisfactory but in need of improvement. Sources of noise and device improvements were identified and suggested.</p> / Doctor of Philosophy (PhD)