Carbon Balance and Metabolic Troubleshooting Using Bioreactor Off-Gas Data: OUR, CER, and RQ in Practice

Carbon Balance Fundamentals

A carbon balance quantifies every gram of carbon entering and leaving a bioreactor, and its closure percentage tells you whether you understand your process. A balance that closes to 95–105% means all major carbon sinks are accounted for. A gap wider than 10% means carbon is going somewhere you are not measuring, and that unknown sink is often the explanation for unexpected process behavior.

The principle is conservation of mass. Carbon enters the bioreactor as substrate (glucose, glycerol, methanol, amino acids in complex feeds) and as dissolved CO₂ from sparging. Carbon leaves as biomass, product, CO₂ in the exhaust gas, dissolved CO₂ in the liquid, organic acid byproducts (acetate, lactate, succinate), and volatile metabolites stripped by aeration (ethanol, acetaldehyde).

The balance equation for a batch or fed-batch fermentation is:

C_{substrate,consumed} = C_biomass + C_CO2 + C_product + C_byproducts ± C_accumulated

Carbon balance closure is the ratio of total carbon out to total carbon in, expressed as a percentage. The target for a well-characterized process is 95–105%. During early development, 90–110% is acceptable because analytical methods may not yet cover all metabolites. Closure below 90% is a red flag requiring systematic investigation.

OUR, CER, and RQ from Off-Gas Data

The oxygen uptake rate (OUR) and carbon dioxide evolution rate (CER) are calculated from the difference between inlet and outlet gas composition, using the nitrogen balance method to account for the change in total gas flow. Since N₂ is metabolically inert, the inlet N₂ flow equals the outlet N₂ flow, which gives the outlet flow rate without needing a direct flow measurement on the exhaust line.

The key equations are:

Where F_in and F_out are the inlet and outlet gas flow rates (typically in mmol/h or L/h at STP), y values are mole fractions measured by the gas analyzer (paramagnetic for O₂, NDIR for CO₂), and V_L is the liquid volume in liters. The mole fraction of N₂ is calculated by difference: y_N2 = 1 − y_O2 − y_CO2 − y_Ar (or 1 − y_O2 − y_CO2 when argon is lumped with nitrogen).

For a detailed walkthrough of setting up the gas analyzers and computing these values step by step, see our companion article on off-gas analysis in fermentation. This article focuses on what to do with OUR, CER, and RQ once you have them: closing the carbon balance and diagnosing metabolic problems.

RQ as a Real-Time Metabolic State Indicator

The respiratory quotient (RQ) reveals the cell's metabolic state in real time, without waiting for offline analytics. An RQ of 1.0 during aerobic growth on glucose means carbon is being fully oxidized through the TCA cycle. Any deviation from 1.0 signals that carbon is being routed through alternative pathways, and the direction and magnitude of the shift tell you which pathway is active.

The practical value of RQ is its response time. While offline metabolite assays take 15–60 minutes (sample, spin, analyze), off-gas analyzers update every 30–60 seconds. An RQ shift from 1.0 to 1.3 in E. coli signals acetate overflow 2–4 hours before acetate accumulation becomes visible in an offline assay. This early warning window is the foundation of RQ-based feed control.

How to Close a Fermentation Carbon Balance

Closing a carbon balance requires measuring or estimating every carbon-containing stream entering and leaving the bioreactor, converting each to moles or grams of carbon, and comparing the totals. The CER from off-gas analysis is the single most important measurement because exhaust CO₂ typically accounts for 40–70% of the carbon leaving a glucose-fed aerobic fermentation.

The carbon content of each component is calculated from its molecular formula:

• Glucose (C₆H₁₂O₆, MW 180): 72/180 = 40.0% carbon by mass
• CO₂ (MW 44): 12/44 = 27.3% carbon
• Biomass: typically ~50% carbon by dry weight (E. coli empirical formula CH_1.77O_0.49N_0.24, MW 25.0, carbon fraction 48%)
• Acetate (C₂H₄O₂, MW 60): 24/60 = 40.0% carbon
• Ethanol (C₂H₆O, MW 46): 24/46 = 52.2% carbon
• Lactate (C₃H₆O₃, MW 90): 36/90 = 40.0% carbon

The cumulative carbon in CO₂ is calculated by integrating CER over time:

C_CO2 (g) = Σ [CER(t) × V_L × Δt] × 12 / 1000

Where CER is in mmol/L/h, V_L in liters, Δt in hours, and 12 g/mol converts from mmol CO₂ to grams of carbon. In practice, trapezoidal integration of the CER time series gives sufficient accuracy.

Common sources of carbon balance gaps

When the carbon balance does not close, the problem is systematic. Work through these sinks in order, because they are ranked by how often they explain the gap:

1. Gas analyzer calibration drift — A 0.1% absolute error in outlet O₂ reading (e.g., 19.9% instead of 20.0%) propagates to a ~5% error in OUR. Calibrate with certified reference gases before and after every run.
2. Dissolved CO₂ not measured — At pH 7.0 and 37 °C, dissolved CO₂ (as HCO₃⁻) can hold 5–15 mmol/L of carbon that never reaches the off-gas analyzer.
3. Volatile byproducts stripped by aeration — Ethanol, acetaldehyde, and other volatiles are carried out by the gas stream but not measured by standard O₂/CO₂ analyzers. This can account for 5–15% of carbon in yeast fermentations.
4. Base neutralization CO₂ — When NaOH neutralizes acetic acid, no CO₂ is produced. But when Na₂CO₃ or NaHCO₃ is used as base, it releases CO₂ that inflates the measured CER.
5. Incorrect biomass composition — Using a generic formula (e.g., CH_1.8O_0.5N_0.2) when the actual composition differs due to lipid accumulation, protein overexpression, or PHB storage can shift the biomass carbon term by 5–10%.
6. Sampling losses — Each sampling event removes 5–20 mL of broth. Over 20–30 samples in a 10 L fermentation, this is 1–6% of the working volume and must be accounted for.

RQ-Guided Metabolic Troubleshooting

RQ deviations from the expected value are the fastest diagnostic signal available to a fermentation operator. The direction and timing of the shift narrow the root cause to one of a few scenarios, each with a specific corrective action.

RQ-Stat Feeding Control

RQ-stat feeding adjusts the substrate feed rate in real time to maintain the respiratory quotient at a target setpoint, typically 1.0–1.05 for E. coli or S. cerevisiae growing on glucose. This prevents overflow metabolism without requiring a pre-calculated exponential feed profile, which makes it robust to batch-to-batch variability in growth rate.

The control logic is simple:

• If RQ > setpoint: decrease feed rate (cells are over-fed, producing overflow metabolites)
• If RQ < setpoint: increase feed rate (cells are under-fed, oxidizing reserves)
• If RQ = setpoint: maintain current feed rate

In practice, the controller uses a proportional or PI algorithm with a dead band of ±0.02–0.05 RQ units to prevent oscillation. Feed rate changes are applied incrementally (5–10% per control cycle, with a cycle time of 1–5 minutes) to avoid sharp metabolic transients.

Sonnleitner and Kappeli (1986) first demonstrated RQ-stat feeding for S. cerevisiae, eliminating ethanol production while maintaining a specific growth rate of 0.3 h⁻¹ in fed-batch. For E. coli, Kim et al. (2004) combined exponential feeding with pH-stat backup and achieved 101 g/L DCW with acetate below 0.5 g/L. Modern implementations use digital controllers that update every 30–60 seconds from the off-gas analyzer signal.

RQ-stat is most valuable when the critical specific growth rate (μ_crit) varies between batches or is not well characterized. For E. coli K-12 strains, μ_crit is 0.20–0.27 h⁻¹; for BL21 strains, it is higher at 0.35–0.45 h⁻¹. An exponential feed designed for K-12 will underfeed BL21, while a feed designed for BL21 will overfeed K-12. RQ-stat adapts automatically to both.

Carbon Balance in CHO Cell Culture

Carbon balance in mammalian cell culture differs from microbial fermentation in three important ways: much lower OUR (0.05–0.5 mmol/L/h vs 5–100 mmol/L/h for E. coli), significant carbon in amino acid feeds, and a large fraction of carbon leaving as lactate rather than CO₂.

In a typical CHO fed-batch producing a monoclonal antibody at 5 g/L, the carbon distribution at harvest (day 14) is approximately:

• CO₂: 40–50% of total carbon out (lower than microbial due to less complete oxidation)
• Biomass: 15–25% (peak VCD ~20 × 10⁶ cells/mL, ~300 pg/cell dry weight)
• Lactate: 5–20% (higher early in culture, often reassimilated after day 5–7)
• Product (mAb): 3–8% (5 g/L mAb ≈ 2.5 g/L carbon)
• Amino acid catabolism: 5–15% (ammonia + organic acid products)
• Unaccounted: 5–15% (improves as analytics mature)

The "unaccounted" fraction is notably higher in CHO than in well-characterized E. coli fermentations. This is partly because chemically defined media contain 20+ amino acids whose catabolic fates are not fully tracked, and partly because CHO cells produce a wider range of organic acid intermediates (citrate, succinate, malate, formate) that may not all be in the analytical panel.

Worked Example: E. coli Fed-Batch Carbon Balance

Frequently Asked Questions

Related Tools

• OUR/CER/RQ Off-Gas Analyzer — Calculate OUR, CER, and RQ from inlet/outlet gas composition using the N₂ balance method.
• OTR & kLa Estimator — Estimate oxygen transfer rate and volumetric mass transfer coefficient for your bioreactor configuration.
• Fed-Batch Calculator — Design exponential, linear, and constant feed profiles for glucose-limited fed-batch fermentation.

References

1. Sonnleitner B. & Käppeli O. (1986). Growth of Saccharomyces cerevisiae is controlled by its limited respiratory capacity: Formulation and verification of a hypothesis. Biotechnology and Bioengineering, 28(6), 927–937. doi:10.1002/bit.260280620
2. Losen M. et al. (2004). Effect of oxygen limitation and medium composition on Escherichia coli fermentation in shake-flask cultures. Biotechnology Progress, 20(4), 1062–1068. doi:10.1021/bp034282t
3. Heinzle E., Biwer A.P. & Cooney C.L. (2006). Development of Sustainable Bioprocesses: Modeling and Assessment. Wiley. doi:10.1002/9780470058916
4. Kim B.S. et al. (2004). High cell density fed-batch cultivation of Escherichia coli using exponential feeding combined with pH-stat. Bioprocess and Biosystems Engineering, 26(3), 147–150. doi:10.1007/s00449-003-0347-8
5. Villadsen J., Nielsen J. & Lidén G. (2011). Bioreaction Engineering Principles, 3rd ed. Springer. doi:10.1007/978-1-4419-9688-6