How oxygen transfer affects biotech fermentation during scale-up

How oxygen transfer affects biotech fermentation during scale-up

A biotech fermentation process that performs reliably in the laboratory can lose productivity at production scale when oxygen delivery fails to keep pace with cellular demand. The challenge extends beyond maintaining a dissolved oxygen setpoint: changes in mixing, broth viscosity, gas dispersion, and the volumetric oxygen transfer coefficient (kLa) can create local oxygen limitations that affect growth, product formation, and batch consistency. Reliable scale-up therefore requires more than reproducing a laboratory control recipe. It requires understanding where oxygen reaches the cells, how quickly they consume it, and what happens when those two rates diverge.

The controlling relationship is oxygen supply versus demand

For a simplified, well-mixed liquid phase, oxygen transfer rate is expressed as:

OTR = kLa × (C* − C_L)

Here, kLa combines the liquid-side mass transfer coefficient with the gas–liquid interfacial area per unit liquid volume. C* is the equilibrium dissolved oxygen concentration under the relevant gas composition, temperature, pressure, and medium conditions. C_L is the actual dissolved oxygen concentration. With consistent units, OTR describes the oxygen transferred into a unit volume of broth per unit time.

Cellular demand is commonly expressed as OUR = qO₂ × X, where qO₂ is the biomass-specific oxygen uptake rate and X is biomass concentration. In a simplified liquid-phase balance, dissolved oxygen declines when OUR exceeds OTR. However, neither biomass nor oxygen demand remains constant throughout fermentation. A feed increase, temperature change, or shift in metabolism can consume oxygen-transfer headroom before the control system catches up.

This relationship also exposes a practical trade-off: maintaining a higher dissolved oxygen concentration reduces the transfer driving force when C* is unchanged. A high setpoint is not automatically safer. The appropriate target must protect process performance without demanding unnecessary agitation, gas flow, or oxygen enrichment.

Why a healthy dissolved oxygen trend can hide trouble

A laboratory vessel often mixes quickly enough for one dissolved oxygen probe to give a useful picture of the culture. A production vessel may contain oxygen-rich regions near gas dispersion zones and oxygen-poor regions farther away. Cells circulate through those regions, experiencing conditions that the installed probe cannot fully describe.

Consider a fed-batch process with concentrated substrate entering near the upper part of the vessel. If feed dispersion is slow, cells near the inlet can encounter elevated substrate concentrations and increased oxygen demand. A probe elsewhere may remain near its setpoint while those cells pass through transient oxygen limitation. Depending on the organism and product, repeated exposure can alter metabolic flux, increase unwanted by-products, or change product formation.

The useful comparison is between mixing time and the timescale of oxygen depletion. If cells consume locally available oxygen faster than mixing replenishes it, a satisfactory bulk reading does not establish satisfactory cellular conditions. Additional probe locations, validated mixing studies, or scale-down experiments can help investigate this gap.

There is no universal critical dissolved oxygen value for biotech fermentation. Sensitivity depends on the organism, medium, production phase, and performance attribute being protected. A threshold established for growth should not automatically be treated as a threshold for product quality.

What changes when the vessel gets larger

Scale-up cannot usually preserve power per volume, impeller tip speed, mixing time, superficial gas velocity, and kLa simultaneously. Choosing which variables to preserve is an engineering decision, not a matter of copying settings.

Agitation illustrates the conflict. Maintaining power per volume in geometrically similar vessels can increase impeller tip speed as scale rises. Restricting tip speed to protect a shear-sensitive culture may instead reduce mixing intensity and gas dispersion. Neither criterion alone guarantees that the production vessel reproduces the laboratory oxygen environment.

Aeration introduces another mismatch. Equal volumes of gas per volume of liquid per minute, commonly called vvm, do not ensure equal superficial gas velocity or gas residence time. Vessel geometry, liquid height, sparger design, and impeller arrangement affect bubble distribution. At excessive gas loading, some impeller systems can lose effective dispersion and pumping performance; increasing airflow may then provide less benefit than expected.

Greater liquid depth also changes pressure. Hydrostatic pressure can increase oxygen solubility toward the bottom of the vessel, while oxygen depletion in rising bubbles changes the gas-side driving force. A single C* value is therefore an approximation in a tall reactor. Temperature, salts, and other medium components further affect oxygen solubility.

Broth behavior deserves particular attention. Filamentous growth, extracellular polymers, and increasing solids can change apparent viscosity during cultivation. Non-Newtonian broth may develop poorly mixed regions even when the initial medium mixed well. Antifoam can also alter bubble coalescence and mass transfer, so its net effect should be measured in the actual formulation rather than assumed.

A kLa value is useful only with its test conditions

Equipment comparisons often reduce oxygen-transfer capability to one kLa number. That number is meaningful only alongside the measurement method, liquid properties, working volume, temperature, pressure, agitation, gas composition, and aeration rate. Clean-water results can screen designs, but they do not establish performance in a viscous production broth.

Dynamic testing and probe response

In a non-respiring liquid, a common dynamic method removes oxygen and measures its recovery after aeration resumes. Fitting the dissolved oxygen response can estimate kLa under suitable assumptions. Sensor response time matters: when the probe is slow relative to oxygen transfer, an uncorrected fit can misrepresent the vessel’s behavior.

Measurements in an active culture are more difficult because respiration continues during the test. Temporarily interrupting aeration may change metabolism, particularly in a high-density culture. Any calculation that assumes constant OUR needs justification, and the perturbation itself must be acceptable for the process.

Off-gas measurements and oxygen balances

Inlet and exhaust gas measurements can support an oxygen balance and estimate OUR. Reliable interpretation requires consistent wet- or dry-gas treatment, calibrated analyzers, accounting for gas-flow changes, and attention to leaks and analyzer delay. Oxygen enrichment makes inlet composition particularly important. During transients, changes in dissolved oxygen inventory mean OTR and OUR should not simply be treated as equal.

The respiratory quotient, derived from carbon dioxide evolution relative to oxygen uptake, can help identify metabolic changes. It is supporting evidence, not a stand-alone diagnosis of oxygen limitation. Its interpretation depends on substrate, organism, and process phase.

For evaluation work, request a performance map rather than a single optimum result. The useful evidence shows how oxygen transfer behaves across expected working volumes, broth conditions, agitation limits, and gas strategies—including late-batch conditions where demand may be highest.

Choosing the intervention without moving the bottleneck

When oxygen supply becomes limiting, the quickest control response is not always the best production strategy. Each intervention changes more than one process variable.

• Increase agitation: potentially improves bubble breakup and mixing, but adds shaft power, heat, and mechanical stress. Motor, gearbox, seal, and cooling limits must be checked.
• Increase airflow: may improve transfer and carbon dioxide removal, but can increase foaming, exhaust-filter loading, and gas-handling demand. Benefits depend on effective dispersion.
• Enrich inlet oxygen: increases the oxygen driving force without requiring the same increase in total gas flow. It adds supply costs and requires appropriate oxygen-service materials, controls, and safety review.
• Increase operating pressure: can improve oxygen solubility, but must remain within vessel and system ratings. Effects on dissolved carbon dioxide and culture performance also need evaluation.
• Reduce or reshape feeding: lowers oxygen demand when substrate supply drives respiration. It may lengthen the batch or change productivity, but can be more reliable than operating continuously at equipment limits.

A dissolved oxygen cascade should make these priorities explicit. Specify actuator order, operating limits, transition behavior, and the response when all available capacity is exhausted. Poor tuning can produce oscillation: agitation rises, oxygen recovers, the controller backs off, and limitation returns. Probe lag and slow mixing can amplify that cycle.

Oxygen enrichment also deserves a separate carbon dioxide check. Reducing total gas flow while maintaining oxygen supply may weaken carbon dioxide stripping. The oxygen trend can improve while another gas-transfer constraint becomes more severe.

What evidence should support a scale-up decision?

There is no universal kLa acceptance limit for all fermentation processes. The defensible requirement is process-specific: demonstrate sufficient oxygen delivery across the intended operating envelope, with justified headroom for demand changes and credible disturbances.

That evidence should connect oxygen-transfer measurements to biological outcomes. Assess expected peak OUR, probe placement and response, feed-location effects, end-of-batch rheology, and utility limits. Where gradients are plausible, scale-down models can expose cultures to alternating oxygen and substrate conditions. Such models need to represent relevant exposure durations, not merely impose an arbitrary low-oxygen period.

For regulated biopharmaceutical production, oxygen-related parameters should be evaluated through process development, risk assessment, qualification, and validation as applicable. Whether dissolved oxygen is a critical process parameter depends on its demonstrated relationship to product quality. A pressure-vessel or safety compliance document does not prove adequate fermentation performance.

This is also where the cross-sector perspective of The Global Industrial Perspective (GIP) becomes useful: reactor performance depends on manufacturing equipment, instrumentation, gas supply, and site energy infrastructure together. A transfer strategy that works technically but exceeds available cooling or oxygen supply is not a deployable scale-up solution.

Before approving the production design, ask for the oxygen-demand profile, the transfer-capacity envelope under representative broth conditions, and the control response to a demand excursion. If those three do not align, a stable laboratory dissolved oxygen trace is not enough reason to proceed.