JMS Paper Introduction

Silicon nanomembranes are ultrathin, free-standing structures that vary from 10 – 100 nm in thickness. These membranes are at least 1 order of magnitude thinner than the thinnest polymeric membranes. Pores can be manufactured on these membranes to have diameters ranging from 5 nm – 10 μm with porosities ranging from 2 – 20%, allowing for a variety of applications (DesOrmeaux). The pore diameter to thickness aspect ratio of these membranes is 1 or greater, which enables improved transport across these ultrathin membranes with lower sample loss and lower transmembrane pressures. Additionally, the ultrathin properties of these membranes allow for both optical and electron transparency, enabling a number of imaging modalities not possible with thicker membranes (Madejski and Lucas). These membranes have previously been shown to have useful properties for a variety of applications ranging from wearable hemodialysis and microvascular barrier mimetics to DNA sensing platforms (Johnson, Smith, Briggs, Salminen, Chung).

Previously, we have described a method for the capture of analytes in microfluidic filtration systems that we termed tangential flow for analyte capture (TFAC) (Dehghani and Lucas, 2019). We demonstrated the ability of this system to capture and release nanoparticles on ultrathin, nanoporous silicon nanomembranes (NPN) with a higher efficiency than on conventional thickness nanoporous polycarbonate track-etch (PCTE) membranes. While this system was ideal for working with bead suspensions, it was not optimized for working with biological solutions, for which filtration processes are often complex and require careful optimization (Bolton, Kozinski, Chan). Biological fluids contain high concentrations of protein (2 – 60 mg/mL) (adachi, marrack) that add complication to the filtration of biofluids, as they will often form thick fouling layers on the membrane surface in most normal flow filtration (NFF) systems. Analyte detection in the presence of these fouling layers is difficult and it is desirable to have a foulant free surface containing only the species of interest. Therefore, we want to determine the critical flux of these systems to optimize capture and detection of our desired analytes.

Critical flux is an important concept in membrane filtration operations. The critical flux determines the system performance for filtering solutions, where it is defined as the flux through the membrane under which no fouling is observed (Field). This is a fine balance between reversible and irreversible fouling that has been well-studied for a variety of polymeric membrane systems in different filtration applications (e.g. reverse osmosis, microfiltration) (Chong, Field). However, there are very few studies that have explored the critical flux properties of micro and nanoscale filtration systems (Debnath, SR ‘19). As systems at this scale become of increasing interest for clinical diagnostic and disease detection applications, there is a need to understand the critical flux and fouling behavior in these devices to better optimize their performance.

In this work, we hypothesize that the lower transmembrane pressures found in ultrathin silicon nanomembrane systems will promote the formation of a flowing cake layer (explain this better). This would be in contrast to the stagnant cake layer that would be expected in higher transmembrane pressure, thick membrane systems. The flowing cake layer will allow the ultrathin silicon nanomembrane systems to operate at higher critical fluxes than thick membrane systems, leading to higher sample throughput. We aim to test this hypothesis by comparing silicon nitride and trach-etch membrane transmembrane pressure in flux-stepping experiments (Zydney). Furthermore, we will demonstrate the development of a computational model in COMSOL Multiphysics that predicts the capture of nanoparticles on ultrathin membranes within the critical flux constraints of the system. We will then confirm the results of the computational model experimentally using gold nanoparticles as a vesicular stand in to show that our model successfully predicts particle capture.