Tangential flow filtration is unavoidably a mechanical process. The feed is pushed across a membrane, recirculated, and pushed again, for hours. For a robust protein that is no problem. For a fragile modality it is a slow accumulation of damage, and AAV capsids, lentiviral envelopes, LNPs, and adenoviral vectors each have different thresholds for how much they can take.
Treating "shear sensitivity" as a single property is the mistake. Each modality fails in its own way, at its own point, through its own weakest structure. Designing a gentle TFF step starts with knowing which failure you are trying to avoid.
Damage in a TFF loop is not one thing. It arrives through several mechanisms, often at once:
| Source | What it does to product |
|---|---|
| Pump shear | The pump is the most aggressive point in the loop. Every pass through it exposes product to high local shear at the tubing, rotor, or diaphragm. |
| Pulsation | Flow that cycles rather than flows steadily swings transmembrane pressure, stressing both product and membrane on every beat. |
| Cavitation | On the suction side, low pressure can form and collapse vapor bubbles, releasing locally destructive energy right at the product. |
| Air-liquid interfaces | Entrained air and foaming expose fragile structures to interfacial forces that denature and aggregate them. |
| Concentration polarization | Product piling up at the membrane wall raises local concentration and stress, and can shear or aggregate sensitive species. |
A single pass through a pump might do little measurable harm. But TFF does not do a single pass. The retentate recirculates back to the feed and goes around again, often hundreds of times over a multi-hour step. The relevant quantity is not the peak shear of one pass but the cumulative dose: shear intensity multiplied by the number of exposures.
This is why a low-shear pump matters more in TFF than almost anywhere else in the process, and why reducing the number of passes (through higher flux or fewer, better-controlled cycles) protects product just as directly as reducing shear per pass.
Different structures break in different ways. The table below is a directional guide to what fails first, not a substitute for characterizing your own material.
| Modality | Vulnerable structure | Dominant failure mode |
|---|---|---|
| AAV | Capsid integrity, genome retention | Relatively robust, but shear and interfaces drive aggregation and full/empty capsid stress |
| Lentivirus | Fragile lipid envelope | Highly shear-sensitive; loses infectious titer quickly as the envelope is disrupted |
| LNP / mRNA | Particle size and encapsulation | Shear and interfacial stress shift size distribution and can leak or expel payload |
| Adenovirus | Non-enveloped capsid | Sturdier than enveloped vectors, but sustained shear still erodes potency at scale |
The key concept: mechanical fragility is not one number. An enveloped vector fails at the membrane and pump long before a non-enveloped one does, and an LNP fails by changing size rather than breaking. The gentle TFF step is the one designed around the most vulnerable structure in your product.
The levers are consistent across modalities, even though the thresholds differ:
| Lever | Why it protects fragile product |
|---|---|
| Low-shear pumping | Cuts the shear delivered on every one of those hundreds of passes. |
| Near-pulseless flow | Holds transmembrane pressure steady, removing the pressure swings that stress product and membrane. |
| Flooded suction | Removes the low-pressure conditions that cause cavitation at the product. |
| Closed, air-free path | Eliminates the air-liquid interfaces that denature and aggregate sensitive species. |
| Fewer, well-controlled passes | Reduces the cumulative dose directly by lowering the exposure count. |
Where Alphinity fits: the TFFi™ system recirculates on the ultra-low-shear, near-pulseless PIXER® diaphragm pump and holds a stable transmembrane pressure, so both the shear per pass and the pressure swings that damage fragile modalities are engineered down from the start.
Several mechanisms at once: pump shear, flow pulsation that swings transmembrane pressure, suction-side cavitation, air-liquid interfaces from entrained air, and concentration polarization at the membrane wall.
TFF recirculates the retentate hundreds of times over a multi-hour step, so the damaging quantity is the cumulative dose, shear per pass multiplied by the number of passes, not the peak shear of a single pass.
Enveloped viral vectors such as lentivirus are the most fragile and lose infectious titer as the envelope is disrupted. LNPs shift size and can lose payload. AAV is relatively robust but can aggregate. Adenovirus is sturdier but still erodes at scale.
Use a low-shear, near-pulseless pump, hold transmembrane pressure stable, flood the suction to prevent cavitation, keep the path closed and air-free, and minimize the number of passes.
Pressure pulsation and TMP stability decide how much of that mechanical damage lands on your product.
Concentrated, fragile feeds are also viscous, and most pumps were not built for that either.
Filtering a fragile modality?
Speak to an engineer