A predictive model of nanoparticle capture on ultrathin nanoporous membranes

Abstract Beyond their use as separation tools, nanoporous membranes can be used for the capture and analysis of biological nanoparticles such as virus or small extracellular vesicles (sEVs). While the challenges of imaging sub-optical particles can be addressed with high resolution imaging techniques, separation challenges remain. We have recently reported on a method of capturing and isolating sEVs, including diagnostically valuable exosomes, in the pores of ultrathin (100 nm thick) nanoporous silicon nitride (NPN) membranes in a microfluidic flow scheme we termed tangential flow for analyte capture (TFAC). Here we present static analytical and dynamic computational models of nanoparticle capture on NPN and compare them to a range of experiments using 60 nm gold nanoparticles (AuNPs) as surrogates for biological nanoparticles. In the presence of 1 mg/mL bovine serum albumin to prevent the formation of gold nanoparticle aggregates, we find that the dynamic model is highly predictive of particle capture over a range of practical flow conditions (∼ 10% error at low ultrafiltration rates, ∼ 1% error at high ultrafiltration rates). Despite the small size of nanoparticles, we find that convective transport dominates diffusive transport during capture (Peclet number > 100), enabling the simple rule-of-thumb that the fraction of nanoparticles captured from solution roughly equals the fraction of the input sample that is ultrafiltered (