How to choose a membrane in a TFF system
Tangential Flow Filtration (TFF) is the workhorse of downstream processing — used to concentrate, purify and exchange the buffer of proteins, vaccines, viral vectors and more. Get the membrane wrong and you lose product, foul the system, or never reach your target. This guide walks a student through the decision, step by step, with animated diagrams.
1. First, what actually happens in TFF?
In dead-end filtration the whole feed is pushed through the filter, so it clogs fast. In tangential (cross) flow, the feed sweeps across the membrane surface. Small molecules and solvent pass through as permeate; anything the membrane holds back stays in the flowing retentate and is recirculated. That sweeping action keeps the surface clean and lets you run for hours.
2. Four questions decide your membrane
Work through these in order. Each one narrows the choice.
① Microfiltration (MF) or Ultrafiltration (UF)?
This is set by what you want to separate. MF membranes are rated by pore size; UF membranes by molecular weight cut-off (MWCO).
Pore size ≈ 0.1 – 0.65 µm. Separates cells & debris from broth (cell harvest / clarification). Product passes through; cells are retained.
MWCO ≈ 1 – 1000 kDa. Concentrates and buffer-exchanges proteins, vaccines, vectors. Product is retained; salts & small molecules pass through.
② Which MWCO (for UF)?
A membrane doesn't have one exact pore — it has a distribution. The golden rule of thumb:
To PASS your product (retain something bigger), pick a MWCO ≥ 3× larger than the product.
Example: to concentrate a 150 kDa antibody, a 30–50 kDa membrane retains it reliably while letting salts/buffer through. Too tight → slow flux & fouling. Too open → product leaks into permeate (yield loss).
③ Which membrane material?
Material decides protein binding (yield), flux, chemical & cleaning robustness.
Very low protein binding → best recovery, low fouling. First choice for sensitive/high-value proteins. Slightly lower flux & narrower chemical range.
High flux, robust, wide pH/chemical tolerance, easy to clean. Some protein binding. A great general-purpose workhorse.
PVDF is rugged and low-binding (often MF/harvest). Match the material to your buffers, cleaning agents (NaOH!) and sterilisation.
④ Which format — flat-sheet cassette or hollow fiber?
High membrane area in a small footprint, high throughput, holds higher pressure. Screened channels give good mixing but more shear — great for robust proteins & large batches.
Open channels = low shear and gentle handling — the choice for cells, viral vectors and shear-sensitive product. Easy to clean; lower area per unit & lower pressure limit.
3. Match it to the job
| Goal | Membrane type | Typical pick |
|---|---|---|
| Cell harvest / clarification | MF | 0.2 µm, hollow fiber (gentle on cells) |
| Concentrate a protein | UF | MWCO ≈ ⅓–⅕ of product, RC or PES cassette |
| Buffer exchange (diafiltration) | UF | Same membrane as concentration; add diafiltration buffer |
| Virus / vector concentration | UF | 100–300 kDa (or 300 kDa–0.05 µm), low-shear hollow fiber |
4. A quick selection checklist
- 1What am I separating, and what is my product's molecular weight / size?
- 2UF or MF? (protein → UF by MWCO; cells → MF by pore size)
- 3MWCO: 3–6× below product to retain it; ≥3× above to pass it.
- 4Material: RC for best recovery/low fouling; PES for flux & robustness.
- 5Format: cassette for area & throughput; hollow fiber for low shear (cells/vectors).
- 6Check chemical compatibility with your buffers & cleaning (e.g. 0.5 M NaOH).
- 7Size the area from your flux (LMH) and time budget — then confirm with a small-scale trial.
5. Common mistakes
- ✕ Choosing a MWCO too close to the product size → product leaks into permeate (yield loss).
- ✕ Using a high-shear cassette on shear-sensitive cells/vectors → damage & low recovery.
- ✕ Ignoring protein binding → a low-cost membrane that quietly adsorbs your product.
- ✕ Running too high a TMP → cake/gel layer, flux collapse and fouling.
- ✕ Skipping the scale-down trial — always confirm flux & recovery before you commit.
Have a TFF question?
Ask the community — get practical answers from people running these systems every day.
Educational overview for students & early process development. Exact cut-offs, materials and areas depend on your molecule, buffers and equipment — always confirm with a scale-down study and your vendor's data.