Date of Graduation


Document Type


Degree Name

Master of Science in Microelectronics-Photonics (MS)

Degree Level



Graduate School


Julie Stenken

Committee Member

Rick Wise

Second Committee Member

Robert Coridan

Third Committee Member

D. Keith Roper


COMSOL, Membrane Separations, Microdialysis, Microfluidics


Microdialysis (µD) sampling is a diffusion-limited sampling method that has been widely used in different biomedical fields for greater than 35 years. Device calibration for in vivo studies is difficult for current non-steady state analytes of interest correlated with both inflammatory response and microbial signaling molecules (QS); which exist in low ng/mL to pg/mL with molecular weights over a wide range of 170 Da to 70 kDa. The primary performance metric, relative recovery (RR), relating the collected sample to the extracellular space concentration varies from 10% to 60% per analyte even under controlled bench-top conditions. Innovations in microdialysis device design have not deviated or improved upon the commercially-available cylindrical geometry for over 35 years. COMSOL Multiphysics finite element method (FEM) software was used to iteratively model and refine microfluidic-based (µF) µD device designs with the primary focus on optimizing channel geometry for improved RR. The primary focus was to improve fluid to membrane perimeter (P) to fluid cross-sectional area (A) and alter the concentration boundary layer (CBL) using passive µF mixing; which are not possible to fabricate using cylindrical geometries. The current µF µD design uses a simple asymmetric linear-looped (LL) geometry optimized with a P/A of 20 vs. 16.4 for a commercial CMA 20 µD probe with an equal 10 mm length. The simulated LL µF µD achieves a 16.1% relative increase in RR vs. experimental data at a 1.0 µL/min inlet flow rate using a 10 kDa FITC-labeled dextran as the analyte. Mixing was implemented and simulated using a modified herringbone geometry (HBM) and compared to an equivalent linear channel (LC). The HBM is shown to shift the CBL and increase diffusive flux at the membrane-fluid interface resulting in a 16.9 ± 0.7% relative increase in RR for 7 flow rates ranging from 0.125 to 2.0 µL/min vs. the LC. The combination of these changes is shown to increase RR above what is currently commercially available.