Transwell Inserts and Permeable Supports: Choosing a Membrane

The insert is the oldest piece of hardware in this hub and still the most used. It is also the one where the specification is fully public, which makes it the best available reference point when a vendor tells you their chip is “equivalent to a Transwell”. Usually it is not, and the published numbers below are how you check.

Two decisions carry almost all of the consequence: membrane material and pore size. Everything else — diameter, plate format, coating — follows from the assay.

Growth areas and volumes by format

Corning’s published figures. These are the numbers to use when normalising anything per unit area, including TEER, permeability and seeding density.

Insert diameter	Plate or dish	Growth area	Medium in the plate well	Medium inside the insert
4.26 mm	96 well	0.143 cm²	0.235 mL	0.075 mL
6.5 mm	24 well	0.33 cm²	0.6 mL	0.1 mL
12 mm	12 well	1.12 cm²	1.5 mL	0.5 mL
24 mm	6 well	4.67 cm²	2.6 mL	1.5 mL
75 mm	100 mm dish	44 cm²	13 mL	9.0 mL

Corning states these values are nominal, may vary with manufacturing variability, and recommends that end users validate their methods independently of the reported values. That is an unusually candid caveat from a vendor and it should be respected: if your assay divides by area, the area is a source of error you did not model.

The volume ratio matters more than it looks. At the 24-well format you are running 0.1 mL apical against 0.6 mL basolateral — a six-fold volume asymmetry across a 0.33 cm² membrane. Compound depletion, metabolite accumulation and pH drift are all asymmetric as a result, and any compound that partitions across the membrane will reach a different steady state in each compartment.

The membrane matrix

Characteristic	Polyester (PET)	Polycarbonate (PC)	PTFE, collagen-coated
Optical properties	Clear	Translucent	Clear when wet
Cell visibility	Good	Poor	Cell outlines only
Tissue culture treated	Yes	Yes	No
Nominal membrane thickness	10 µm	10 µm	30 µm
Matrix/ECM coatable	Yes	Yes	Yes
Collagen treated	No	No	Yes
Available pore sizes	0.4, 3.0, 0.8 µm	0.4, 3.0, 5.0, 8.0, 12.0 µm	0.4, 3.0 µm

Corning notes that membrane manufacturing processes are inherently variable and actual thickness values may vary up to 60% of nominal depending on membrane type and measurement technique. A “10 µm membrane” is therefore a class, not a dimension. If your model depends on diffusion distance through the support, that 60% is in your error budget whether you account for it or not.

The practical rule: if you need to see the cells, use PET. If you need the widest pore range or the highest pore density, use PC and accept that you are imaging through a translucent membrane. PTFE-collagen is a specialist choice for cultures that need the collagen surface, and Corning describes its coating process as covering every fibril of the filter matrix while retaining porosity, in contrast to conventional coating that produces an occluding film layer.

Pore density is the number people forget

Pore size gets specified. Pore density almost never does, and it determines how much of the membrane is actually open.

Pore size	Polycarbonate nominal density	Polyester nominal density
0.4 µm	1 × 10⁸ pores/cm²	4 × 10⁶ pores/cm²
1.0 µm	—	1.6 × 10⁶ pores/cm²
3.0 µm	2 × 10⁶ pores/cm²	2 × 10⁶ pores/cm²
5.0 µm	4 × 10⁵ pores/cm²	—
8.0 µm	1 × 10⁵ pores/cm²	1 × 10⁵ pores/cm²
12.0 µm	1 × 10⁵ pores/cm²	—

At 0.4 µm the polycarbonate membrane has 25 times the pore density of the polyester one. Two membranes with the same stated pore size have wildly different open area, and therefore different blank resistance, different diffusive flux and different apparent permeability. If you are comparing a published permeability value against your own and you used a different membrane material at the same pore size, you have not run the same experiment.

Corning also notes that PTFE has no defined pore density at all, because it is a tortuous-path membrane rather than a track-etched one. It does not have pores in the same sense.

Choosing a pore size

Corning’s own guidance, which matches the literature:

0.4 or 3.0 µm — drug transport studies. Small pores keep cells on their side of the membrane.
3.0 µm or larger — cell invasion, chemotaxis and motility. Cells must be able to get through.
Migration will not occur with pores smaller than 3.0 µm. That is a hard floor, not a tendency.

The complication Corning states explicitly is that the ability of cells to migrate through pores depends on the cell line and culture conditions as well as pore size, and they recommend testing a range of pore sizes with appropriate controls for critical experiments. In practice that means an invasion assay ported from one cell type to another is a new assay, and the pore size is one of the variables to re-derive.

The rest of the family

The Transwell line is four products with different jobs, and the wrong one is frequently bought.

Standard Transwell insert. Hanging design that keeps the membrane about a millimetre off the bottom of the well, with a patented self-centring feature that prevents medium wicking between the insert and the well wall, and windows in the sides giving access to the lower compartment. The one-millimetre standoff exists so a co-cultured monolayer at the bottom of the well is not scratched when the insert is moved.
Snapwell. A 12 mm tissue-culture-treated polycarbonate or clear polyester membrane on a detachable ring. The ring is the whole point: it lets the membrane be transferred into an Ussing-type chamber or a diffusion cell. If your endpoint is a chamber measurement, buying a standard insert and trying to cut it out later is a bad afternoon.
Netwell. Polystyrene inserts with polyester mesh bottoms for 6- and 12-well plates, intended for 3D culture in gels and for tissue explants and microcarrier-grown cells. Mesh, not a track-etched membrane — a different device for a different job.
HTS Transwell-96. Ninety-six inserts in a single robotics-friendly unit, with a stated well bottom elevation of 1.3 mm. This is the format that goes on a liquid handler; the individual inserts are not automation parts.

TEER: the number, and why yours is not comparable to theirs

Transepithelial electrical resistance is the standard integrity readout on an insert, and it is the most commonly mis-reported figure in barrier biology. The mechanics are simple; the caveats are not.

Normalisation. Resistance is inversely proportional to culture area, because a larger area offers more parallel paths for the current. The convention is therefore to subtract a blank and multiply by the membrane area:

TEER = (R − R_blank) × A_membrane, in Ω·cm²

Which is why the growth-area table at the top of this page is load-bearing. Use the area of the insert you actually used, not the one in the paper you are comparing against.

Electrode configuration changes the answer. Holzreuter and Segerink’s review of TEER in organs-on-chips lays out the mechanism: manually placed chopstick electrodes deliver a non-uniform electrical current density, so different regions of the barrier contribute unequally to the measured impedance. Fixed circular electrodes in an EndOhm-style chamber create a more uniform field and remove errors caused by electrode placement, which is why the same monolayer gives a different number on the two setups.

The instruments are more alike than the numbers suggest. The EVOM2 applies an AC square-wave current of 10 µA amplitude at 12.5 Hz with 1 Ω resolution; the Millicell ERS uses the same 10 µA square wave at 12.5 Hz, also at 1 Ω resolution. The newer EVOM3 offers adjustable fixed currents of 2, 4 or 10 µA, auto-ranges from 1 Ω to 100,000 Ω, resolves to 0.1 Ω below 200 Ω and averages 20 samples per reading. So the meters agree on the physics. The spread between laboratories comes from electrode geometry, placement, temperature and blank handling — not from the box.

Confluence sensitivity is brutal. The same review cites Odijk and colleagues showing that a drop in cell coverage from 100% to 99.6% — one missing cell in a monolayer — causes an 80% decrease in measured TEER. TEER is not a graded measure of barrier quality across that range. It is close to a binary detector for holes, which makes it excellent for catching a failed monolayer and poor for ranking two good ones.

Temperature and medium confound it. Medium resistivity varies with temperature, and for long channels with small cross-section the medium contribution can overshadow the biological signal entirely. Impedance spectroscopy handles this by extracting medium resistance from the spectrum rather than from a separate blank, giving an intrinsic temperature control; single-frequency Ohm’s-law meters do not.

What to record, every time. Electrode type and geometry, applied current and frequency, temperature at measurement, blank value and how it was obtained, membrane material, pore size, pore density and area. A TEER value without those is a number, not a measurement.

When the insert is the wrong device

Three honest cases:

You need flow. An insert is static. If shear stress on the apical surface is part of the biology, you are choosing between a channel device and a perfused plate — see perfusion and flow control.
You need to see the barrier form in real time under flow. Membrane-free platforms exist specifically to remove the support from the optical path; MIMETAS builds the OrganoPlate around surface-tension patterning with no artificial membrane, and reads TEER with a dedicated instrument rather than chopsticks.
You need throughput above 96 with automated liquid handling. The HTS-96 format is the ceiling for inserts. Above that you are on a plate-based platform, and the comparison to make first is organ-on-chip versus Transwell versus static culture.

How we can help

We do not sell inserts and you do not need us to buy them — they are catalogue items from multiple distributors. Where we are useful is the step after: when the insert format does not exist for the geometry you need, when you want a membrane in a custom device and need a fabricator who will bond one properly, or when you are trying to reconcile a TEER number from a paper against your own bench. Send us the specification and we will tell you whether it is a catalogue purchase or a fabrication problem.

Transwell inserts and permeable supports: pore size, membrane material and the TEER number