Bioreactor gas mixing systems

Oxygen is almost insoluble in culture medium. At 37 °C, water in equilibrium with air holds roughly 7 mg per litre, and Henry’s law fixes that ceiling: raising the oxygen fraction in the gas raises the saturation value proportionally and does nothing else. A high-density culture consumes its entire dissolved inventory in seconds. A bioreactor therefore does not store oxygen, it delivers it continuously, and the whole design problem is a rate problem.

What the sparger is actually setting

Oxygen transfer rate is the product of a coefficient and a driving force: OTR = kLa × (C* − C_L). The driving force is bounded by solubility and by the dissolved-oxygen setpoint the cells need. Everything else is kLa — a lumped term combining the liquid-film transfer coefficient with the gas–liquid interfacial area per unit volume. Gas flow appears in kLa only through superficial gas velocity and gas hold-up; it is not the variable being controlled. A sparger that halves bubble diameter at the same volumetric flow doubles interfacial area, which is why microspargers with sintered or laser-drilled orifices buy far more transfer than turning the rotameter up.

Bubble size is not, however, free to choose. In electrolyte-rich media, coalescence behaviour changes: dissolved salts suppress bubble coalescence, so the same sparger produces a different steady-state bubble population in medium than in water, and a kLa measured by water gassing-out overstates the real value. Antifoam does the opposite — it lowers surface tension and promotes coalescence, and a dose sufficient to control foam can measurably depress kLa. This is a routine and under-recognised cause of a culture becoming oxygen-limited without any parameter changing on the control screen.

Where mammalian cells are actually damaged

Animal cells in suspension survive impeller shear at industrial tip speeds far better than early literature assumed. What kills them is bubble rupture at the free surface: as a bubble bursts, the collapsing film generates local energy dissipation orders of magnitude above the bulk average, and cells adsorbed to the bubble interface are destroyed. Pluronic F-68 works by adsorbing to that interface and preventing cell attachment to it — a surface-chemistry protection, not a viscosity effect. Small bubbles, ideal for transfer area, have long residence times and high surface-to-volume ratio, so the same choice that improves oxygenation worsens interfacial damage and foaming.

The constraint that decides scale-up

Carbon dioxide is the mirror image of oxygen and the reason large vessels fail in ways small ones do not. CO2 is highly soluble, so it is removed only by stripping into the gas phase, and stripping depends on gas volume passed per unit liquid volume. Scale-up conventionally holds power per volume or tip speed constant, which forces volumetric gas flow per volume down as the tank grows, while hydrostatic head at the sparger raises dissolved CO2 further. Accumulated dissolved CO2 acidifies the medium, base addition compensates and drives osmolality up, and the combination — not oxygen starvation — is a common cause of a process that performed at two litres losing productivity at two thousand.

Gas blending hardware exists to serve this: mass flow controllers proportioning air, oxygen, nitrogen and carbon dioxide, cascaded from the dissolved-oxygen probe and, for CO2, used both to acidify and to be deliberately withheld so the sweep can strip. Constant kLa, constant P/V and constant volumetric gas flow cannot all be preserved across scales. Choosing which to sacrifice is the design decision.