Publication Date


Document Type


First Advisor

Tan, Jifu

Degree Name

M.S. (Master of Science)

Legacy Department

Department of Mechanical Engineering


Isolation of circulating tumor cells (CTCs) from patient derived blood samples for cancerdiagnostics and monitoring is challenging due to the extremely low CTC concentration. This thesis work evaluates the performance of passive microfluidics devices with different designs based on computational modeling. A new cell separation method is proposed based on velocity differentiation of varying cell sizes. All numerical studies were conducted in 2D based on a fluid structure interaction code where the fluid flow was solved based on the lattice Boltzmann method (LBM) implemented in Palabos, and the cell dynamics were simulated based on the molecular dynamics package LAMMPS, with both being coupled through the immersed boundary method (IBM). 280 unique simulations were conducted on the millimeter length scale of microfluidic devices, owing to the efficient 2D mode running on parallel computers. The simulation framework was validated by closely reproducing the results of particle separation by a pinched flow fractionation device from the literature, and by comparing separation behavior with experimental work.

To investigate the efficacy of different designs, five different device layouts were simulated with 25 different cell parameter combinations of varying bending stiffness and sizes,each simulation containing 34 cells. Designs considered were disordered hyperuniform (HU), deterministic lateral displacement (DLD), staggered, grid, and hexagonal micropost layouts. Efficacy of devices was estimated by computing normalized exposure time (NET), defined as the ratio of transport time when a cell is close to any micropost over the total transport time in the device, for each cell flowing in each device. In addition, each cell’s trajectory deviation from corresponding streamline was also calculated by comparing the trajectories of cells against streamlines of undisturbed flow (without cells) using the L2 norm of the deviation in the direction perpendicular to flow direction, denoted as SD. Both NET and SD were characterized over different device layouts, cell sizes, and cell bending stiffnesses. Simulation results showed that cell diameters have a much stronger effect than cell bending stiffness on exposure time for all devices. Specifically, for cell size of 30 µm, the DLD device was found to have the best cell exposure time, NET = 25.1%. The HU layout was second best with NET = 17.2%, while the staggered layout produced NET = 12.0%. Both hexagonal and grid layouts performed very poorly in comparison, producing NET = 1.9% and NET = 3.2% respectively.

In addition, the proposed cell separation method based on cell size was confirmed experimentally by obtaining velocities of separate PC3 (average diameter d = 20.0 µm) and CEM(d = 12.3 µm) cell populations. Average velocities of 536.9 µm/s for PC3 and 397 µm/s for CEM were recorded, yielding a 35.2% difference. Using identical geometry over a range of cell diameters, the computational model produced an estimate of a 24.7% difference in velocities. Further, the effect of geometric parameters on the obtained velocity differentiation was studied on both grid and hexagonal micropost layouts. The most important parameter affecting cell velocity differentiation was the gap distance between microposts perpendicular to the flow direction, denoted as the pinched length Pl. In all cases, smaller cells (d ≤ 10 µm) had much higher velocity variations, which depended on the alignment of cells in reference to rows of microposts. The combined results indicate that DLD devices may be more effective for cell capture applications over hyperuniform device, and that other devices with regular layouts perform poorly. However, devices with a regular layout can separate cells effectively based on the cell velocity differentiation technique.


65 pages




Northern Illinois University

Rights Statement

In Copyright

Rights Statement 2

NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.

Media Type