Fig. 8.6. Foam sensing and control unit.
vices which have been manufactured include horizontal rotating shafts, centrifugal separators and jets spraying on to deflector plates (Hall et al., 1973; Viesturs et al., 1982). Unfortunately most of these devices have to be used in conjunction with an antifoam.
A load cell offers a convenient method of determining the weight of a fermenter or feed vessel. This is done by placing compression load cells in or at the foot of the vessel supports. When designing the support system for a fermenter or other vessel, the weight of which is to be measured by load cells, the principle of the three-legged stool should be remembered. Three feet will always rest in stable equilibrium even though the supporting surface is uneven. If more feet are provided, the additional feet must each be fitted with means of adjustment or precision packing to ensure load bearing on all the feet.
A load cell is essentially an elastic body, usually a solid or tubular steel cylinder, the compressive strain of which under axial load may be measured by a series of electrical resistance strain gauges which are cemented to the surface of the cylinder. The load cell is assembled in a suitable housing with electrical cable connecting points. The cell is calibrated by measuring compressive strain over the appropriate range of loading. Changes of resistance with strain which are proportional to load are determined by appropriate electrical apparatus.
It is therefore possible to use appropriately sized load cells to monitor feed rates from medium reservoirs, acid and base utilization for pH control and the use of antifoam for foam control. The change in weight in a known time interval can be used indirectly as a measure of liquid flow rates.
Real-time estimation of microbial biomass in a fermenter is an obvious requirement, yet it has proved very difficult to develop a satisfactory sensor. Most monitoring has been done indirectly by dry weight samples (made quicker with microwave ovens), cell density (spectrophotometers), cell numbers (Coulter counters) or by the use of gateway sensors which will be discussed later in this chapter. Other alternative approaches are real-time estimation of a cell component which remains at a constant concentration, such as nicotinamide adenine dinucleotide (NAD), by fluo-rimetry or measurement of a cell property which is proportional to the concentration of viable cells, such as radio frequency capacitance.
It is well established that fluorimetric measurements are very specific and rapid, but their use in fermentation studies is limited. The measurement of NAD, provided that it remains at a constant concentration in cells, would be an ideal indirect method for continuous measurement of microbial biomass. In pioneer studies, Harrison and Chance (1970) used a fluorescence technique to determine NAD-NADH levels inside microbial cells growing in continuous culture. Einsele et al (1978) mounted a fluorimeter on a fermenter observation port located beneath the culture surface which enabled the measurement of NADH fluorescence in situ, making it possible to determine bulk mixing times in the broth and to follow glucose uptake by monitoring NADH levels. Beyeler et al. (1981) were able to develop a small sterilizable probe for fitting into a fermenter to monitor NADH, which had high specificity, high sensitivity, high stability and could be calibrated in situ. In batch culture of Candida tropicalis, the NAD(P)H-dependent fluorescence signal correlated well with biomass, so that it could be used for on-line estimation of biomass. Changes in the growth conditions, such as substrate exhaustion or the absence of oxygen, were also very quickly detected.
Schneckenburger et al. (1985) used this technique to study the growth of methanogenic bacteria in anaerobic fermentations. They thought cost was a problem, fluorescence equipment being too expensive for routine biotechnology applications when the minimum price was about US$10,000. Ingold (Switzerland) have developed the Fluorosensor, a probe which can be integrated with a small computer or any data transformation device (Gary, Meier and Ludwig, 1988).
Dielectric spectroscopy can be used on-line to monitor biomass. Details of the theory and principles of this technique have been described by Kell (1987). At low radio frequencies (0.1 to 1.0 MHz), a microbial cell membrane will act as a capacitor, and become charged by the so-called /^-dispersion effect (Schwan, 1957), making it possible to discriminate between microbial cells, gas bubbles and insoluble media particles. The size of this /3-dispersion is linearly proportional to the membrane enclosed volume fraction up to high cell densities. Kell et al. (1987) were able to show that the capacitance (dielectric permittivity) was linearly proportional to the biomass concentration. The output for unicellular organisms is proportional to the mean cell radius whereas with mycelial suspensions the output remains linear for increases in biomass of a particular cell morphology. The capacitance has been shown to give a linear response with biomass using a number of strains of bacteria, yeasts, mycelial fungi, plant and animal cells.
The sterilizable probe can be inserted directly into a fermenter using a 25-mm diameter port. Fouling of the gold electrodes in the probe can be avoided by the automatic application of electrolytic cleaning pulses. The sensor ('Bug meter') is manufactured by Aber Instruments (Aberystwyth, Wales) and marketed by Applikon (Schiedam, The Netherlands). It has a capacitance range of 0.1 to 200 pF (picoFarads), which is equivalent to 0.1 to 200 mg dry weight cm"3 (approximately 106 to 2 X 109 cells cm"3 of Saccha-romyces cerivisiae). The resolution depends on the type of cells and the conductivity of the medium, but is normally 0.1 mg dry weight cm"3. In order for the sensor to work effectively the suspending medium must have a minimum conductance. Yeast slurries after acid washing (Chapter 6) are satisfactory, but before such washing there may be a need for extra salts in the medium in order to make measurements. This sensor has proved ideal for yeast cells and is now being used by the brewing industry to control yeast pitching rates (Boulton et al, 1989).
In most aerobic fermentations it is essential to ensure that the dissolved oxygen concentration does not fall below a specified minimal level. Since the 1970s steam sterilizable oxygen electrodes have become avail
Electrolyte Glass wool
Cathode fath0"" (Ag spiral) ,pt'
Electrolyte Glass wool
Cathode fath0"" (Ag spiral) ,pt'
Fig. 8.7. Construction of dissolved-oxygen electrodes: (a) galvanic, (b) Polarographie (Lee and Tsao, 1979).
able for this monitoring (Fig. 8.7). Details of electrodes are given by Lee and Tsao (1979).
These electrodes measure the partial pressure of the dissolved oxygen and not the dissolved oxygen concentration. Thus at equilibrium, the probe signal of an electrode will be determined by:
F(02) = C(02)X?t where P(02) is the partial pressure of dissolved oxygen sensed by the probe, C(Oz) is the volume or mole fraction of oxygen in the gas phase, Pr is the total pressure. The actual reading is normally expressed as percentage saturation with air at atmospheric pressure, so that 100% dissolved oxygen means a partial pressure of approximately 160 mmHg.
Pressure changes can have a significant effect on readings. If the total pressure of the gas equilibrating with the fermentation broth varies, the electrode reading will change even though there is no change in the gas composition. Changes in atmospheric pressure can often cause 5% changes and back pressure due to the exit filters can also cause increases in readings. Allowance must also be made for temperature. The output from an electrode increases by approximately 2.5% per °C at a given oxygen tension. This effect is due mainly to increases in permeability in the electrode membrane. Many electrodes have built-in temperature sensors which allow automatic compensation of the output signal. It is also important to remember that the solubility of oxygen in aqueous media is influenced by the composition. Thus, water at 25°C and 760 mmHg pressure saturated with air will contain 8.4 mg 02 dm~3, while 25% NaCl in identical conditions will have an oxygen solubility of 2.0 mg 02 dm"3. However, the measured partial pressure outputs for 02 would be the same even though the oxygen concentrations would be very different. Therefore it is best to calibrate the electrode in percentage oxygen saturation. More details on oxygen electrodes and their calibration has been given by Hailing (1990).
In small fermenters (1 dm3), the commonest electrodes are galvanic and have a lead anode, silver cathode and employ potassium hydroxide, chloride, bicarbonate or acetate as an electrolyte. The sensing tip of the electrode is a teflon, polyethylene or polystyrene membrane which allows passage of the gas phase so that an equilibrium is established between the gas phases inside and outside the electrode. Because of the relatively slow movement of oxygen across the membrane, this type of electrode has a slow response of the order of 60 seconds to achieve a 90% reading of true value (Johnson et al, 1964). Buhler and Ingold (1976) quote 50 seconds for 98% response for a later version. These electrodes are therefore suitable for monitoring very slow changes in oxygen concentration and are normally chosen because of their compact size and relatively low cost. Unfortunately, this type of electrode is very sensitive to temperature fluctuations, which should be compensated for by using a thermistor circuit. The electrodes also have a limited life because of corrosion of the anode.
Polarographic electrodes, which are bulkier than galvanic electrodes, are more commonly used in pilot and production fermenters, needing instrument ports of 12, 19 or 25 mm diameter. Removable ones need a 25 mm port. They have silver anodes which are negatively polarized with respect to reference cathodes of platinum or gold, using aqueous potassium chloride as the electrolyte. Response times of 0.05 to 15 seconds to achieve a 95% reading have been reported (Lee and Tsao, 1979). The electrodes which can be very precise may be both pressure and temperature compensated. Although a polarographic electrode may initially cost 600% more than the galvanic equivalent, the maintenance costs are considerably lower as only the membrane should need replacing.
Prototypes of a fast response phase fluorometric sterilizable oxygen sensor are now being developed (Bambot et al., 1994). The sensor utilizes the differential quenching of a fluorescence lifetime of a chromo-phore, tris(4,7-diphenyl-l,10-phenanthroline)rutheni-um(II) complex, in response to the partial pressure of oxygen. The fluorescence of this complex is quenched by oxygen molecules resulting in a reduction of fluorescence lifetime. Thus, it is possible to obtain a correlation between fluorescence lifetime and the partial pressure of oxygen. However, at room temperature when a Clark-type oxygen electrode shows a linear calibration the optical sensor shows a hyperbolic response. The sensitivity of the optical sensor when compared with an oxygen electrode is significantly higher at low oxygen tensions whereas the sensitivity is low at high oxygen tensions. The sensor is autoclavable, free of maintenance requirements, stable over long periods and gives reliable measurements of low oxygen tensions in dense microbial cultures.
Dissolved oxygen concentrations may also be determined by a tubing method, described by Phillips and Johnson (1961) and Roberts and Shepherd (1968). The probe consists of a coil of permeable teflon or propylene tubing within the fermenter through which is passed a stream of helium or nitrogen. The oxygen, which diffuses from the fermentation medium through the tubing wall into the inert gas stream, is then determined using a paramagnetic gas analyser (see next section). Times of 2 to 10 minutes are required before taking readings. The tubing will withstand repeated sterilization and has been used continuously for up to 1000 hours at pilot scale.
If it is necessary to increase the dissolved oxygen concentration in a medium this may be achieved by increasing the air flow rate or the rpm of the impeller or a combination of both processes. Another way is to increase the ratio of oxygen to nitrogen in the input gas using a variable proportionating valve while maintaining a constant gas flow rate (Siegall and Gaden, 1962). The cost of this technique would normally restrict its use to laboratories and pilot plants.
Inlet and exit-gas analysis
Was this article helpful?