Computational modeling and fluorescence microscopy characterization of a two-phase magnetophoretic microsystem for continuous-flow blood detoxification

Abstract

Magnetic beads can be functionalized to capture and separate target pathogens from blood for extracorporeal detoxification. The beads can be magnetically separated from a blood stream and collected into a coflowing buffer solution using a two-phase liquid-liquid continuous-flow microfluidic device in the presence of an external field. However, device design and process optimization, i.e. high bead recovery with minimum blood loss or dilution remain a substantial technological challenge. We introduce a CFD-based Eulerian-Lagrangian computational model that enables the rational design and optimization of such systems. The model takes into account dominant magnetic and hydrodynamic forces on the beads as well as coupled bead-fluid interactions. Fluid flow (Navier-Stokes equations) and mass transfer (Fick's law) between the coflowing fluids are solved numerically, while the magnetic force on the beads is predicted using analytical methods. The model is demonstrated via application to a prototype device and used to predict key performance metrics; degree of bead separation, flow patterns, and mass transfer, i.e. blood diffusion to the buffer phase. The impact of different process variables and parameters - flow rates, bead and magnet dimensions and fluid viscosities - on both bead recovery and blood loss or dilution is quantified for the first time. The performance of the prototype device is characterized using fluorescence microscopy and the experimental results are found to match theoretical predictions within an absolute error of 15%. While the model is demonstrated here for analysis of a detoxification device, it can be readily adapted to a broad range of magnetically-enabled microfluidic applications, e.g. bioseparation, sorting and sensing.

Continuous-Flow Separation of Magnetic Particles from Biofluids: How Does the Microdevice Geometry Determine the Separation Performance?

“…The particles are assumed spherical with a diameter of 4.9 µm and a density equal to 2000 kg•m −3 . The saturation magnetization and susceptibility values employed are M s,p = 1.5 × 10 4 A•m −1 and χ p,e = 0.25, which are within the range of magnetization and susceptibilities reported for commercial beads [27,41]. The surrounding fluids are considered non-magnetic due to their low susceptibilities in comparison to the beads.…”

“…The computational model employed in this work has been presented in our previous publications and has also been experimentally validated [22,27]. Briefly, the model involves an Eulerian-Lagrangian approach to predict the bead movement under the influence of a magnetic field; the beads are considered as discrete elements and their trajectories are computed as follows:…”

“…These studies have analyzed the influence of multiple variables and parameters on the separator performance, such as particle size, flow rates, magnet dimensions and positions, fluid properties, etc. [4,[25][26][27][28]. However, a realistic analysis of the geometrical features of microdevices is missing, and literature data on the effect of the geometry of microfluidic channels on particle magnetophoresis are scarce.…”

“…As such, the non-symmetrical, U-shaped (U) chip sections derived from wet etching might influence the system performance when compared to channels with the strictly rectangular (R) sections that arise from lithography techniques [36][37][38][39]. Indeed, in our previous work [27], we addressed the separation of magnetic beads from blood solutions in both glass and SU-8 microchannels fabricated with different techniques, and we observed that, even when the same inlet velocities and magnets were applied, the systems rendered different particle recovery. This was attributed to the distinct chips cross sections as well as to the different channel thicknesses (as a result of applying different bonding techniques).…”

“…More specifically, different Y-Y shaped microfluidic devices are tested by assessing the effect of both channel length and cross-sectional shape and thickness on the separation of magnetic beads for a model biofluid (i.e., blood) and their recovery into an aqueous solution with a permanent magnet. For that purpose, we used our experimentally validated CFD model comprised of a fluidic solver that is coupled to a customized magnetic model [27]. The model is employed to accurately describe the separation of beads inside various channel geometries under different flow rate conditions while keeping invariant the magnet position with regard to the channel surface and the channel width (i.e., the depth is determined by the total chip thickness, which varies with the fabrication/bonding method).…”