Back in November we were visited by Tom Berkelman of BioRad to see if he could use our NPN membranes to successfully separate some proprietary nanoparticles from IgG in a solvent with secret ingredients. These particles have an interesting absorbance spectrum as well as strong fluorescence.
During that visit and in subsequent separations in the McGrath lab and at BioRad, we clog the membrane after passing only 20-30 µL. This occurs at concentrations of 0.1 mg/mL and higher, when using the constant pressure setup. Lower concentrations seem to allow greater total volume to pass through the filter, but clogging still occurs. We want to find a way to use our membranes to filter solutions with particles/protein at higher concentrations.
One hope we have of addressing this clogging is a “reverse” centrifuge technique. When Tom originally visited us on 11-20-13, he made mention of the Sartorius Centrisart
The centrifugal force acts to equilibrate the solution levels inside and outside the inner tube. This drives solution through the filter while also driving larger particulates away from the filter towards the bottom of the centrifuge tube. In this configuration, blocked particles should be less likely to stay at the filter and clog flow.
Jim suggested we try a similar approach using our Gen2 SepCon assemblies. The attached figure below
Gen2 Sepcon placed into 2 mL tube
shows the Gen2 being used with a 2 mL centrifuge tube. We tried Gen2 because it has a narrower profile and allows for more volume (~150 µL) around the side than the Gen3 tubes.
I started by testing a mixture of 500 nm yellow fluorospheres and ~20 nm red fluorospheres in 1X PBS. I diluted each stock 100 to 1 by volume. I believe this corresponds to 3 E9 particles/mL of 500 nm and 5 E13 particles/mL of 20 nm. This mixture has typically blocked flow through a membrane after ~30 µL of flow. I started the centrifuge at 2krpm for a series of 1 and then 5 min runs. There was very little if any flow visible. Maybe 20 µl. I upped the rpms to 3 k for 5 mins, saw a little flow. Then added 5 min at 4 krpm. At this point approximately ~100 µl of filtrate had been collected. Next I increased the rpms to 5k with continued success for 10 minutes. The membrane broke at 6 krpm. Little to no red or yellow particles seemed to have made it through the filter, as seen in the attached fluorescence plots.
I replicated the above test using two mixtures of materials Tom B. left after his visit. BR1 is combination of 1500 µl PBS/fairy dust + 15 µl 10mg/ml IgG + 150 µl “T24”.
BR2 is 1500 µl PBS/fairy dust + 15 µl “25b”.
“T24” is a particle prep that creates particles with ~32nm diameter at 122 ppm.
25b is “unfractionated IgG conjugation reaction” antibody to particle ratio 2:1. 524 ppm particles, 1.05 mg/ml IgG.
Both had 100 ul or more pass through the membrane to the 20+ 500 nm solution discussed above. I’ve attached the measured absorbance and flourescence data showing little to no particles passing through the membrane while at least some IgG has passed through. No IgG seems to have passed through in the BR2 test. (Or the concentration may simply be too low to measure using the nanoquant plate.)
BR 1 absorbance showing T24 absorbance peaks at ~325 and ~460 nm.
BR 1 fluorescence
BR 2 absorbance
BR 2 fluorescence
I think these positive results merit further study. Immediate actions might be to modify the current 2 ml centrifuge tubes so that less total volume is required. Currently we need at least 1.5 ml. If we fill the bottom of the tube with PDMS for example, we might only need 0.5 ml.
We should also repeat these experiments with higher concentrations of IgG and particles.
(In a parallel effort, we recently purchased some Labview microfluidics equipments to drive filtration with a syringe pump while monitoring the pressure building up behind the membrane.)
I have a couple questions: Does anyone know of a stock of open bottom pieces for the Gen2 assemblies? I have many of the closed versions, which are not ideal for this test.
I believe I need to use 100 um gasket material for the Gen2 SepCons. Correct? These are 300 um chips.
Some time before I noticed that placing the membrane in the metal deposition chamber at slight angle to the incoming metal atoms stream makes thin metal layer only on one side of the pore, shadowing the rest of it. Rotating the membrane sample after the metal layer was deposited on the front and making sure…
In this post I show the sieving coefficients of simulations with a thin membrane, a thick membrane, or a no membrane case as they approach equilibrium. The simulations were carried out for 48 hours, and plotted for 1 hour time points. Both thin and thick membrane had the same pore distribution – monodisperse 10 nm…
Introduction As a continuation of my rotation project and a beginning to my PhD work in the McGrath lab, I wanted to finalize my shear stress calculations within our blood vessel microfluidic device. All previous shear stress calculations were purely analytical, with many assumptions. This work aimed to minimize those assumptions while providing some publishable…
Xingyu et al; Supplement The authors of this submission to ACS Nano apply techniques for the creation of mesoporous silica thin films with vertically aligned nanochannels to create ultrathin (< 100 nm) silica membranes with ~ 2 nm nanochannels. The technique involves the growth of silica around surfactant templates followed by the removal of the…
This is an update of the research on the on-chip EO pump. Due to a mistake in determining the direction of particle flow, it now looks like the particles in this post were moving by electrophoresis rather than electroosmosis. The particles were negatively charged (-40 mV zeta potential) because of a COOH group on the…
Last week I made a post showing the diffusion of DDM and BSA through SC267. I an effort to increase the negative charge associated with the membrane material I proposed using UV-Ozone exposure. A while back there was some concern that the UV-Ozone system was contaminated by PDMS, and that a standard cleaning protocol needed…
This could be a breakthrough for SepCon applied to biologicals and concentrated nanoparticles.
The role of the g-force in keeping the membrane clear of particles is interesting. We need to spin high enough pull the cake away. A little counter-intuitive, but spinning harder might be the key to prevent clogging.
Chris recently asked about the relative number of particles in solution and pores in the membrane.
I calculate a typical 5 slot sepcon has ~ 1 E8 million pores.
The solution with 20 nm NPs used above has 5 E13 NPs/mL.
At these concentrations, there are many more particles than pores.
This could be a breakthrough for SepCon applied to biologicals and concentrated nanoparticles.
The role of the g-force in keeping the membrane clear of particles is interesting. We need to spin high enough pull the cake away. A little counter-intuitive, but spinning harder might be the key to prevent clogging.
Chris recently asked about the relative number of particles in solution and pores in the membrane.
I calculate a typical 5 slot sepcon has ~ 1 E8 million pores.
The solution with 20 nm NPs used above has 5 E13 NPs/mL.
At these concentrations, there are many more particles than pores.