Insight & Intelligence

More »


More »
September 01, 2018 (Vol. 38, No. 15)

pH Sensor Recalibration Based on Exhaust CO2 Concentration

An Alternative Procedure for Bioprocess Transfer across Plants, Sites, and Scales

  • The pH of the culture medium is a critical process parameter in cell culture bioprocesses. To successfully transfer processes between bioprocess systems and sites (i.e., in the course of process development and scale up) it is essential to ensure that pH readings in the different systems are comparable. In cell culture bioprocesses, pH sensors are often recalibrated after sterilization based on offline pH measurements. This can have pitfalls because factors such as sensor age, temperature, and CO2 degassing can influence the measurements, making direct cross-site comparisons of pH values difficult.

    In this tutorial, we describe a method for accurate inline pH sensor recalibration based on the CO2 concentration in the exhaust, which was developed by Roche Pharma Biotech Penzberg. This method allows matching of the starting pH of carbonate-buffered systems in process transfers, scale up, and scale down across plants, sites, and scales. A potentially flawed sample-based pH offline reading is no longer necessary, decreasing the risk of pH-derived comparability issues.

  • pH Sensor (Re)Calibration

    Click Image To Enlarge +
    Figure 1. Workflow for pH sensor calibration and recalibration. The pH sensor is calibrated outside the bioreactor using buffers at pH 4 and pH 7 (left). The sensor is then built into the bioreactor and sterilized by autoclaving (middle). Afterward, the sensor is recalibrated to correct for changes caused by autoclaving (right). Routinely, recalibration is based on offline pH readings (right, top). However, this method is prone to errors. Alternatively, the pH sensor can be recalibrated based on the CO2 concentration in the exhaust (right, bottom).

    In cell culture bioprocessing, pH sensor calibration usually follows this workflow (Figure 1): The pH sensor is calibrated outside the bioreactor and then built into the bioreactor for sterilization. After autoclaving, the bioreactor is filled with sterile medium. Bioprocess engineers now routinely recalibrate the sensor to correct for possible changes caused by autoclaving. They take a sample of medium from the bioreactor, measure its pH offline with a standard pH meter or blood gas analyzer, and use the offline reading for sensor recalibration. Direct comparability of bioreactor pH sensor readings between systems and sites cannot be ensured, however, because the offline pH measurements (and therefore the recalibration) are prone to error.

    The time between sampling and offline measurement can cause temperature shifts and CO2 degassing, leading to a pH value of the sample that no longer reflects the value of the medium inside the bioreactor. Furthermore, sensor age, response time, daily adjustment procedures, and other factors may influence offline pH readings. pH offsets of up to 0.3 between sites and plants have been shown to exist, and they cannot be detected using the equipment which caused them in the first place. This makes process transfer and troubleshooting difficult.

    Bioprocess engineers from Roche Pharma Technical Development developed an alternative recalibration procedure, which is performed inline without sampling and therefore avoids the possible errors of offline pH measurements. The pH of cell-free, carbonate-buffered culture medium is indirectly determined by measuring the CO2 concentration in the bioreactor exhaust and using this data for pH sensor recalibration (Figure 1).

  • Measuring Principle

    According to the Henderson-Hasselbalch equation, CO2 concentration in the gas phase must be identical in multiple systems, if the media, temperature, pressure, and pH are identical. At equilibrium, the net CO2 mass transfer between the gas phase and the liquid phase is zero. In this state, the CO2 concentration in the gas phase is not a function of parameters that influence mass transfer kinetics, and can therefore be considered scale independent. Thus, one can indirectly infer the culture medium pH from the CO2 concentration in the exhaust and use this information for pH sensor recalibration.

  • Correlation of Exhaust CO2 Concentration and Medium pH

    Click Image To Enlarge +
    Figure 2. (A) Measurements were carried out in four glass bioreactors with a maximum working volume of 2 L. The CO2 concentration in the exhaust was measured at different medium pH values. The four bioreactors were operated under different experimental conditions. (Factors such as the working volume, agitation speed, P/V, and kLa were varied.) (B) pH and exhaust CO2 concentration were measured in a bioreactor with a working volume of 100 L, a bioreactor with a working volume of 400 L, and in two 2-L bioreactors.

    Four 2-L glass bioreactors were filled with cell-free culture medium. The working volumes, agitation speed, gassing (percentage of CO2, VVM), kLa, power number per volume (P/V), and the presence or absence of baffles were varied.

    The CO2 concentration in the exhaust was quantified with a DASGIP® GA4 exhaust analyzer. The device uses robust dual-beam infrared CO2 sensor technology by BlueSens. Four CO2 analyzer channels allow the precise exhaust analysis of four bioreactors in parallel with a single DASGIP GA4. Integrated sensors allow for an automated compensation of pressure, humidity, and temperature effects.

    To correlate the CO2 concentration in the exhaust with the pH of the medium, the latter was measured online using the bioreactors’ pH sensor. The pH of the cell-free culture medium was quantified using the internal pH sensors, which were lowered into the bioreactors under nonsterile conditions via the lid. This approach allows verification of bioreactor sensor signals without sampling and offline measurement, although it cannot be applied under sterile conditions. The correlation of exhaust CO2 and medium pH are described by a calibration curve that was then used in further experiments for inline recalibration of the pH sensor. Figure 2A shows the correlation of the exhaust CO2 concentration and the pH of the cell-free culture medium. Parameters like the agitation speed, P/V, kLa, and the presence or absence of baffles varied, but they did not influence the correlation.

  • Verification of Scale Independence of the pH Sensor Recalibration Procedure

    To verify the method for inline pH sensor recalibration, the process was scaled up. The experimental procedure described previously was applied in bioreactors with working volumes of 100 L and 400 L. For the 100-L and 400-L bioreactors, it was ensured by installing a bypass, so that the gas flow to the DASGIP GA4 did not exceed 250 sL/h. Parameters like P/V, kLa, tip speed, gassing, and stirrer configuration were not kept constant. The pH of the cell-free culture medium was measured using the internal pH sensors, which were inserted under nonsterile conditions, as described previously. Figure 2B shows that the correlation between the two parameters was scale independent.

  • Verification of the Recalibration Method by Measuring Lactate Levels

    Click Image To Enlarge +
    Figure 3. CHO cell culture processes were performed. The pH sensors were recalibrated based on the CO2 concentration in the exhaust (exhaust-based), or based on offline readings (standard).

    It is well known that secretion of lactate by CHO cells via a lactate-proton symporter increases with increasing pH of the culture medium. Therefore, the lactate concentration in the medium can serve as a reliable indirect signal of the medium’s pH. CHO cell bioprocesses were performed in glass bioreactors with a working volume of 2 L. The cells were cultivated for up to 200 hours. Two different pH setpoints were used. Up to eight replicates were analyzed per experimental condition. Before inoculation, the pH sensors were either recalibrated based on the CO2 concentration in the exhaust or using a standard procedure based on offline pH readings. All other process parameters were the same for all runs until 200 hours into the batch. After that, the feed strategy was modified.

    In the runs in which the pH sensors were recalibrated based on offline measurements, the lactate concentrations varied (Figure 3, blue). In the runs in which the pH sensors were recalibrated based on the CO2 concentration in the exhaust, the lactate concentrations in the replicates remained virtually identical (Figure 3, green and red). This indicates that sensors recalibrated based on the exhaust CO2 concentration deliver more consistent measurements than sensors recalibrated using standard procedures based on offline measurements.

  • Conclusion

    When the exhaust-related pH measurement method is used, pH comparability can be established and pH excluded as a source of potential deviation. Process development and troubleshooting can be conducted more efficiently. In summary, the exhaust-based pH reference method provides a simple, cost-effective, reproducible, and robust way to detect otherwise undetectable but relevant pH offsets in carbonate-buffered systems.

Related content