# Bioprocessing Scale-Up: Single-Use Bioreactors (SUB) vs. Stainless Steel Stirred Tanks for mAbs and Recombinant Proteins

# Quantitative Transport Phenomena, Oxygen Mass Transfer ($k_L a$), Kolmogorov Shear Dissipation, and Techno-Economics in Modern Biomanufacturing

# Executive Summary

The global biopharmaceutical industry has undergone a monumental structural pivot toward high-titer Monoclonal Antibodies (mAbs), Antibody-Drug Conjugates (ADCs), bispecific antibodies, and recombinant therapeutic enzymes. Modern fed-batch Chinese Hamster Ovary (CHO) cell cultures routinely reach viable cell densities ($\text{VCD}$) exceeding $20\text{–}35 \times 10^6\text{ cells/mL}$ and expression titers of $4\text{ to }8\text{ g/L}$.

As biological titers have climbed, facility engineering has transformed. The historical standard of massive $10,000\text{–}20,000\text{ L}$ stainless steel (SS) bioreactor facilities is increasingly challenged by modular $500\text{–}2,000\text{ L}$ Single-Use Bioreactors (SUB) utilizing gamma-irradiated polymer bag liners.

However, scaling mammalian cell culture is governed by delicate, highly constrained transport physics:

1. Oxygen Mass Transfer Bottlenecks: Meeting elevated Oxygen Uptake Rates ($OUR$) requires volumetric mass transfer coefficients ($k_L a \ge 25\text{ h}^{-1}$) without inducing high-velocity bubble shearing.
2. Kolmogorov Shear Stress vs. Cell Fragility: Unlike robust bacteria (E. coli), wall-less mammalian cells are susceptible to hydrodynamic shear. The local Kolmogorov eddy length scale ($\eta_k$) must never approach the eukaryotic cell diameter ($12\text{–}20\mu\text{m}$).
3. Dissolved Carbon Dioxide ($pCO_2$) Toxicity: While sparging pure oxygen satisfies $DO$, inadequate gas velocity fails to strip metabolic $CO_2$. Accumulation of $pCO_2 > 120\text{–}150\text{ mmHg}$ drives medium acidification, increases base addition, elevates osmotic pressure ($>380\text{ mOsm/kg}$), and severely degrades antibody glycosylation profiles.

This technical guide provides the fundamental bio-transport derivations, hardware hydrodynamic comparisons, and a fully worked industrial scale-up case study from a 50 L seed bioreactor to a 2,000 L commercial Single-Use Bioreactor (SUB).

# 1. Gas-Liquid Oxygen Mass Transfer: The $k_L a$ Transport Model

In aerobic cellular metabolism, oxygen is sparingly soluble in aqueous nutrient broths ($C^* \approx 0.21\text{ mmol/L}$ at $37^\circ\text{C}$ in ambient air). The bioreactor must continuously deliver oxygen across the gas-liquid interface to match the metabolic consumption of the culture:

Where:

• $\text{OTR}$: Oxygen Transfer Rate ($\text{mmol/L}\cdot\text{h}$)
• $\text{OUR}$: Oxygen Uptake Rate ($\text{mmol/L}\cdot\text{h}$)
• $k_L a$: Volumetric gas-liquid mass transfer coefficient ($\text{h}^{-1}$)
• $C^*$: Saturated dissolved oxygen concentration at operating temperature and pressure ($\text{mmol/L}$)
• $C_L$: Bulk dissolved oxygen concentration in liquid broth ($\text{mmol/L}$, typically maintained at $30\text{–}50\%$ air saturation)
• $q_{\text{O}_2}$: Specific cellular oxygen consumption rate ($\approx 0.25\text{ to }0.45\text{ pmol/cell}\cdot\text{h}$ for CHO)
• $X_v$: Viable cell density ($\text{cells/mL}$)

         Gas Sparger Bubble (Pure O2 / Air)
                       │
                       ▼
       ┌───────────────────────────────┐
       │ Gas Film Resistance (Neglig.) │
       ├───────────────────────────────┤ ◄── Gas-Liquid Interface
       │ Liquid Film Resistance (1/kL) │ ◄── Rate-Limiting Barrier (Film Thickness δ)
       ├───────────────────────────────┤
       │ Bulk Nutrient Medium (CL)     │
       └───────────────┬───────────────┘
                       ▼
            Mammalian CHO Cell (OUR = qO2 · Xv)

# 1.1 The Van 't Riet Correlation for Bioreactors

The volumetric mass transfer coefficient is parameterized by agitator power dissipation per unit volume ($P/V$) and superficial sparge gas velocity ($u_g$):

For non-coalescing biological media containing surfactants (such as Pluronic F-68 / Poloxamer 188), typical exponents are:

# 1.2 Bubble Dynamics & Specific Interfacial Area ($a$)

The volumetric mass transfer coefficient $k_L a$ is the product of the liquid-film mass transfer coefficient ($k_L$) and the specific interfacial surface area ($a$):

$a = \frac{6 \cdot \phi_{gas}}{d_{32}}$

Where:

• $\phi_{gas}$: Gas phase holdup fraction ($\text{m}^3\text{ gas / m}^3\text{ liquid}$, typically $0.02 - 0.08$)
• $d_{32}$: Sauter mean bubble diameter ($\text{m}$)

The Sauter mean bubble diameter is governed by the equilibrium between turbulent shear breakup and bubble coalescence:

$d_{32} = C_b \cdot \left( \frac{\sigma_L^{0.6}}{(P/V)^{0.4} \cdot \rho_L^{0.2}} \right) \cdot \left( \frac{\rho_L}{\rho_G} \right)^{0.2}$

• Microspargers ($20\text{–}40\mu\text{m}$ frits): Generate tiny bubbles ($d_{32} \approx 0.5\text{–}1.0\text{ mm}$), creating an immense interfacial area ($a > 250\text{ m}^2/\text{m}^3$) that achieves high $k_L a$ at very modest gas flow rates.
• Macrospargers ($1.0\text{–}2.0\text{ mm}$ drilled holes): Generate larger bubbles ($d_{32} \approx 3\text{–}5\text{ mm}$), which minimize surface cell disruption upon bursting and provide high volumetric buoyancy for stripping volatile metabolic byproducts.

# 2. Shear Stress & The Kolmogorov Microscale of Turbulence

A persistent myth in bioprocess scale-up is that "impeller tip speed kills mammalian cells." In reality, empirical hydrodynamic research confirms that bulk laminar shear in stirred tanks is rarely lethal to cells protected by Pluronic surfactants.

Cell lysis occurs predominantly in two specific physical regimes:

1. The Bubble Disruption / Cavitation Zone at the Surface: As sparged gas bubbles burst at the liquid-gas headspace interface, the collapsing fluid jet exerts localized shear stresses exceeding $100\text{–}300\text{ Pa}$, shredding adjacent cell membranes.
2. Microscale Turbulent Eddy Dissipation: When energy dissipation rates ($\varepsilon_{turb} = P / (\rho V)$) in the impeller discharge stream become extreme, the smallest turbulent eddies become smaller than the cell diameter.

# 2.1 The Kolmogorov Microscale Equation

Where:

• $\eta_k$: Kolmogorov eddy length scale ($\text{m}$)
• $\nu$: Kinematic broth viscosity ($\approx 1.0 \times 10^{-6}\text{ m}^2/\text{s}$)
• $\varepsilon_{max}$: Maximum local turbulent energy dissipation rate ($\text{W/kg}$)

   Eddy Size (ηk) >> Cell Diameter (d_cell ≈ 15 µm)  ──► Safe: Cells Carried Along Streamlines
   Eddy Size (ηk) ≈ Cell Diameter (d_cell ≈ 15 µm)  ──► LETHAL: Turbulent Energy Shreds Membrane!

# 3. Dissolved Carbon Dioxide ($pCO_2$) Stripping Dynamics

While oxygen transfer is driven by high concentration gradients, carbon dioxide removal is governed by fluid liquid-to-gas stripping:

In small-scale laboratory bioreactors ($3\text{–}10\text{ L}$), the headspace surface-area-to-volume ratio ($A_{surface}/V$) is large, allowing metabolic $CO_2$ to desorb effortlessly.

At commercial scale ($2,000\text{ L}$ SUB or $12,000\text{ L}$ SS):

• $A_{surface}/V$ collapses by over $80\%$.
• Hydrostatic liquid height ($H_L \approx 1.8\text{–}3.5\text{ m}$) increases the bottom partial pressure of $CO_2$.
• If operators rely solely on pure oxygen sparging at low total gas flow ($u_g < 0.002\text{ m/s}$) to meet $DO$, $pCO_2$ rapidly exceeds $140\text{ mmHg}$.

High pCO2 (>140 mmHg) ──► Acidifies Medium (pH Drops) ──► Auto-Addition of NaOH / Na2CO3
                                                                   │
                                                                   ▼
Glycosylation Defects & Early Apoptosis ◄── Elevated Osmolality (>380 mOsm/kg)

# 3.1 Gas-Liquid Transfer Equivalence for Carbon Dioxide

Because both oxygen absorption and carbon dioxide desorption occur across the same boundary layer film, the mass transfer coefficient for $CO_2$ is scaled directly by molecular diffusivity via penetration theory:

$k_L a_{\text{CO}_2} = \sqrt{\frac{D_{\text{CO}_2}}{D_{\text{O}_2}}} \cdot k_L a_{\text{O}_2} \approx 0.91 \times k_L a_{\text{O}_2}$

However, while oxygen transfer benefits from high concentration gradients (driving force $\Delta C \approx C^* - C_L$), $CO_2$ stripping is strictly volumetric-flow limited:
$\text{Mass Stripped} = \dot{V}_{gas} \cdot \frac{P_{\text{CO}_2}}{R \cdot T}$
If the total sparge gas flow rate (ballast air) is insufficient ($<0.05\text{ vvm}$), metabolic $CO_2$ accumulates regardless of agitator speed.

# 4. Hardware Comparison: Single-Use Bioreactor (SUB) vs. Stainless Steel (SS)

       Single-Use Bioreactor (SUB)                      Stainless Steel Bioreactor (SS)
┌──────────────────────────────────────┐          ┌──────────────────────────────────────┐
│ • Pre-sterilized Multi-layer Bag    │          │ • Electropolished 316L Stainless     │
│ • Bottom-mounted Angled Impeller    │          │ • Dual/Triple Rushton/Hydrofoil      │
│ • Microsparger + Drilled Tube Sparge │          │ • Fixed Ring Sparger                 │
│ • Zero Cleaning Validation Required │          │ • Full CIP/SIP Steam Distribution    │
│ • Volume Capped at 2,000 - 6,000 L  │          │ • Scalable to 20,000 L Commercial    │
└──────────────────────────────────────┘          └──────────────────────────────────────┘

# 5. Worked Industrial Scale-Up Case Study: Scaling a mAb Process to 2,000 L SUB

# 5.1 Process Baseline in Seed Bioreactor

• Cell Line: Recombinant CHO-DG44 producing humanized IgG1 mAb.
• Peak Viable Cell Density: $X_{v,max} = 28 \times 10^6\text{ cells/mL}$.
• Specific Oxygen Consumption ($q_{\text{O}_2}$): $0.32\text{ pmol/cell}\cdot\text{h} = 3.2 \times 10^{-10}\text{ mmol/cell}\cdot\text{h}$.
• Maximum Required $\text{OUR}$:

• Dissolved Oxygen Setpoint: $40\%$ air saturation ($C_L = 0.40 \times 0.21\text{ mmol/L} = 0.084\text{ mmol/L}$).
• Equilibrium Oxygen Concentration (Pure $\text{O}_2$ enriched): $C^* \approx 0.52\text{ mmol/L}$.

# 5.2 Calculating Required Volumetric Mass Transfer ($k_L a_{req}$)

Adding a $25\%$ control safety margin:

# 5.3 Hardware Design for 2,000 L Single-Use Vessel

• Working Volume ($V_w$): $2,000\text{ L} = 2.0\text{ m}^3$.
• Vessel Diameter ($T$): $1.30\text{ m}$ ($A_{col} = 1.327\text{ m}^2$).
• Liquid Height ($H_L$): $1.51\text{ m}$ ($H/T = 1.16$).
• Impeller Configuration: Dual 3-blade pitched hydrofoil (Elephant Ear style, down-pumping), diameter $D = 0.45\text{ m}$ ($D/T = 0.346$).
• Impeller Power Number ($N_p$): $1.45$.

# Step 1: Agitator Speed Sizing for Constant Power per Volume

Targeting a conservative specific power dissipation $P/V = 45\text{ W/m}^3$ (to ensure $\eta_k > 22\mu\text{m}$):

# Step 2: Tip Speed Verification

# Step 3: Kolmogorov Microscale Calculation

Assuming peak local dissipation in the impeller discharge zone is $\varepsilon_{max} \approx 12 \times \varepsilon_{avg} = 0.529\text{ W/kg}$:

Result: Since $\eta_k = 37.1\mu\text{m}$ is significantly larger than the CHO cell diameter ($15\mu\text{m}$), zero shear-induced cell lysis will occur.

# Step 4: Sparging Strategy (Dual Sparger Architecture)

To decouple oxygen transfer from $CO_2$ stripping:

• Microsparger ($20\text{–}40\mu\text{m}$ porous frit): Delivers fine oxygen microbubbles dedicated exclusively to meeting $k_L a_{req}$ at low gas flow rates, minimizing surface bubble burst damage.
• Drilled-Hole Macrosparger ($1.0\text{ mm}$ orifices): Delivers ballast air ($0.05\text{–}0.10\text{ vvm} = 100\text{–}200\text{ standard L/min}$) generating larger bubbles that sweep up the liquid column, stripping dissolved $CO_2$ and maintaining $pCO_2 \le 90\text{ mmHg}$.

# 6. Bioprocess Scale-Up Checklist

1. Antioxidant Additive Screening (bDtBPP): Perform gamma-irradiation stability testing on single-use polyethylene contact films. Bis(2,4-di-tert-butylphenyl)phosphate (bDtBPP) leachables inhibit CHO growth at concentrations above $0.1\text{ mg/L}$.
2. Kinetics of Base Addition (Plume Mixing): At $2,000\text{ L}$ scale, blend time ($\tau_{blend}$) stretches to $35\text{–}50\text{ seconds}$. Feed concentrated base ($0.5\text{–}1.0\text{ M NaOH}$) directly into the high-velocity discharge stream of the lower impeller to prevent localized alkaline cell killing ($\text{pH} > 8.5$).
3. Antifoam Optimization: Excessive silicone emulsion antifoam coats the gas-liquid interface, reducing $k_L a$ by up to $50\%$. Size the bioreactor headspace with a mechanical foam breaker or acoustic resonance defoamer.