Dev. Chem. Eng. Mineral Process. 13(1/2), pp. 63-70, 200.5. Cost Optical Sensor for Online Measurement of Biomass M. He’, X.M. Li2, G.F. Qin’, S.K. Nguang’* and X.D. Chen’ ’Department of Electrical and Electronic Engineering Department of Chemical and Materials Engineering University of Auckland, Private Bag 92019, Auckland, New Zealand ~~ ~~~ ~ A low cost optical on-line biomass sensing system for fermentation processes is presented in this paper. The calibration curve of the sensor obtained by using cell cultures in tap water revealed a linear profile in the range 0.1-4.0 g/L. The sensor was shown to have a fast response time and was insensitive to the natural light. A good reproducibility was obtained between the optical sensor and cell dry-weight method values. Introduction Modem bioprocesses are the most complex of all the fields of process engineering due to their high dimension, non-linearity and dynamical characteristics. The ability to control bioprocesses at their optimal states accurately and automatically is of considerable interest to many fermentation industries. Thls enables them to reduce their production costs and increase the yield, while at the same time maintaining the quality of the metabolic product. However, it should be noted that designing a control system for bioprocesses is not a straightforward task due to: (i) significant model uncertainty; (ii) lack of reliable on-line sensors which can accurately detect the important state variables; (iii) non-linear and time varying nature of the system; and (iv) slow response of the process, in particular for cell and metabolic concentrations. This work is mainly focused on the second problem. Over several years, the key state variables (such as the concentration of biomass) in a fermentation process are commonly determined using off-line laboratory assays with a long measurement delay [I]. This limits the range of control algorithms that can be applied to the process. * Author for correspondence (firstname.lastname@example.org). 63 M.He, X.M. Li. G.F. Qin, S.K. Nguang and X.D.Chen Optical measuring technology is widely used not only in remote control and microelectronic technology but also in biotechnology. Most of the online measuring techniques for cell densities and concentrations are based on optical methods of turbidimetry, fluorescence, capacitance or resistance of the microbiological culture. With turbidimetric methods, light transmittance and/or scattering in the media can be detected by different mechanisms, e.g. transmission (detecting lower turbidity values), forward scattering, backscattering or lasers . By controlling the cells with various dyes injected in the media, the fluorescence detector sensor can be used to monitor different cellular components. The sensors based on the capacitance measurement use the principle of measuring the dielectric permittivity of microbial suspensions . In t h s paper, a low-cost infrared optical sensor was constructed for continuous on-line measurement of biomass concentrations. The calibration curve was obtained using cell cultures in tap water, and presented a linear profile in the range 0.1-4.0 g/l corresponding to 0.5-4.0 V output signal. The sensor was shown to have a fast response time and was insensitive to the natural light. A good reproducibility was obtained between the optical sensor and cell dry-weight method values. Beer's Law Beer's Law states that the absorbance of a solution is dependent on three factors: (a) the molar absorptivity, the value of which depends on the absorbing species and on the wavelength used; (b) the path length of the solution through which the light must past; and (c) the concentration of the solution. Therefore: A = Ebc where A is the absorbance; & is the constant of proportionality (called the molar absorptivity); b is the path length of the sample, i.e. the distance of a beam of monochromatic radiation passing through the cuvette; c is the concentration of the material in solution. Different molecules absorb different radiation wavelengths. Transmittance is closely related to absorbance, and the efficiency of transmitting in the materials depends on the physical characteristics of the material . Although the fermentation solution consists of different materials which absorb different wavelengths, it is still possible to find a monochromatic radiation that is suitable for the yeast with a particle size of about 1 pm. Design Considerations The main principle of this low-cost optical sensor design is to utilize Beer's Law and suitable detector circuits for measuring the biomass on-line. (0 Circuit design The design of the low-cost optical sensor is illustrated in Figure 1. The emitter and the detector are placed in the solution. The light falling on the detector decreases as the biomass concentration increases. This leads to a drop in voltage at the negative 64 Cost Optical Sensor for Online Measurement of Biomass input terminal of the operational amplifier. This drop in voltage forces the operational amplifier to increase the current through the emitter and, hence, bring the voltage at the negative input terminal back to the reference voltage. The final output voltage of the circuit also increases. Hence, we refer to it as the Light Feedback Balance Bridge (LFBB). The concentration of the biomass is converted into an analogue signal through the LFBB circuit of the sensor. This analogue signal is then converted into a digital signal by an A/D converter. A computer receives the digital data via a serial port. All the data are displayed, analysed and stored in the computer. Balance Bridge 1 AID Converter 1 Computer Displayer Figure 1. Light Feedback Balance Bridge. Figure 2. Laboratory probe. 65 M. He, X.M. Li, G.F. Qin, S.K.Nguang and X D. Chen (ii) Probe design With commercial applications in mind, probes are designed to fit the existing port of a standard fermentor. The structures of the probes are illustrated in Figures 2 and 3. The laboratory probe has a wider range of measurements, and the industry probe is more robust. Features of the Laboratory Probe (see Figure 2): The distance between the emitter and the detector can be adjusted from both ends. The top-end adjustment enables determination of the electrical sensitivity range for different mediums. The bottomend adjustment allows adjustment of the dynamic response range and the sensitivity of the probe in the solution. Figure 3. Industry probe. Features of the Industry Probe (see Figure 3): The emitter and the detector are tightly sealed so that they can be cleaned and disinfected together with the fermentor. The light transmitting distance is fixed and only suited for a particular biomass measurement. Results and Discussion The experiments were performed at the Biotechnology Laboratory, the Department of Chemical and Materials Engineering, the University of Auckland, New Zealand. A. Equipment, Materials and Weight Methods Equipment: Laboratory probe, power supply (Topward 6303A), digital multimeter (Agilent 34401A), Lambda 35 UVNIS Spectrometer (Perkin-Elmer Instruments), scale (Mettler PL300), shaker, stirring machine, fermentation container. Microorganism: Saccharomyces cerevisiae (dried baker's yeast packed for Goodman Fielder Milling & Baking N.Z. Ltd). 66 Cost Optical Sensorfor Online Measurement of Biomass Medium: YEPD medium used in the fermentation experiments is the composition of yeast extract (10 gA), peptone (20 gA), dextrose (20 g/l) and commercial antifoam ( 10 dropdl). Pre-weight method: Weigh the dry yeast first and then pour it into the water. Use shaker or stirring machine to make the solution uniform. Dry-weight method: First centrihge the broth sample of 10 mL for 10 minutes at 4000 rpm. Then measure the dry weight of yeast after dehydrating in the centrifuge tubes at 65°Cfor 48 hours [ 5 ] . Figure 4. Experimental bench. B. Caliiwation Curve Figure 4 shows the experimental set-up. According to our requirements, different concentrations of the solutions were manual mixed. The probe was placed in the solution which was consequently stirred by a magnetic stirring machine. The data was collected by the sensor. The following tests were performed. Pure biomass solution test: Between l g and 8g of dried baker's yeast were added to 200 ml of tap water in the 250 ml flask, then well mixed by the magnetic stirrer. The probe was placed in the solution and the sensor collected the data. The cell concentration was determined by the dry weight method. Nutrimental biomass solution test: YEPD medium (200 ml) was added to the 250 ml flask. After sterilizing at 110°C for 30 minutes, between Ig and 8g of dried baker's yeast were added. The optical sensor and the dry weight method were used to measure the cell concentration. 67 M. He, XM.Li, G.F. Qin, S.K.Nguang and X D. Chen C. On-line Biomass Measurement 1. Shaker culture: A series of mediums in 250 ml flasks with volumes from 80 ml to 200 ml and glucose concentrations from 20 gfl to 50 gfl were 2. prepared. After sterilizing at 110°C for 30 minutes, 1 g of dried baker's yeast was added to each flask. They were cultured in the shaker at 30°C and 200 rpm for 2 to 8 hours. Then 20 .ml of concentrated glucose solution (300 g/l) were fed to each flask every hour, and one of the flasks was taken out as a sample for determining its cell concentration by both the optical sensor and the dry weight method. Small-scale bioreactor culture: Two litres of medium was added to the 3-litre bioreactor (New Brunswick Scientific Co., Inc.) with a working volume of 2.5 litres. The complete assembly with the optical probe was sterilized by autoclaving at 1 10°C for 30 minutes. Then 4 g of dried baker's yeast were inoculated into the bioreactor. The fermentation process was operated for 4 or 5 hours. The fermentation broth was sampled every half hour, and the cell concentration was also determined by the dry weight method. Results and Discussion From the experiments, it was found that different monochromatic radiation wavelengths possessed similar features after passing through the solution. Thus, a near-infrared phototransistor was chosen that is less sensitive to the natural light. Further experiments showed that the emitting diode SE5455 with the phototransistor SD5443 were the most robust pair. They can work at 935 nm wavelength and 125°C temperature, and are also insensitive to the natural light and the colour of the solution. As shown in Figure 5, the longer the distance between the emitter and detector then the larger the range of the output voltage. The best distance between the emitter and the detector in our experiments was 10 c m ,..... c 6 0.5 0 50 + Series1 8 Series2 100 Distance between emitter and detector (mm) Figure 5. Transmission distance effects on circuit output. 68 Cost Optical Sensorfor Online Measurement of Biomass calibration 6 2 -Linear 0 0.0 1.0 2.0 3.0 4.0 5.0 concentration (g/L) Figure 6. Pure biomass test results. Figure 6 shows the calibration curve for the pure biomass solution. The output voltage from the sensor shows a linear relationship with the concentration of the biomass. Figure 7 shows the test results for the nutrimental biomass solution, it also shows a linear relationship between the output voltage of the sensor and the biomass concentration. The three series depicted in Figure 7 represent three different experiments completed at the three different times. The reproducibility of the sensor was shown to be very good. Series2 3.5 0.5- + Series 1 n ’. 31 A - 0 0 1 2 3 4 5 concentration Figure 7. Nutrimental biomass test results. 69 M.He, X.M. Li, G.F. Qin, S.K.Nguang andX.D. Chen In Figure 8, the on-line biomass concentration measured by the sensor was compared to the biomass concentration obtained by the dry-cell weight method. The voltages from the sensor are converted to biomass concentrations according to the calibration curve. A linear relationship was also obtained in thls test. Dynamic experiment 0 0.5 1 1.5 2 Biomass, Dry cell weight (glL) Figure 8. Online biomass measurements. Conclusions This paper has proposed a novel way of designing a low-cost infrared sensor for online measurement of biomass concentrations. This infrared sensor has been successfully used for measuring biomass concentration ranges from 0.1 to 4.0 gA with good reliability. The sensor has been shown to have a fast response time and is insensitive to the natural light. The cost of the sensor is less than US$SO. This robust and low-cost infrared sensor may provide an attractive solution for the fermentation industries and dairy industries. References 1. 2. 3. 4. 5. 70 Olsson, L., and Nielsen, J. 1997. On-line and in situ monitoring of biomass in submerged cultivations, Trends in Biotech, 15(12), 517-522. MacMichael, G.,Armiger, W.B., Lee, J.F., and Mutharasan, R. 1987. On-line measurement of hybridoma growth by culture fluorescence, Biorechnol. Technol., 1,213-21 8 . Salgado, A.M., Folly, R.O.M., and Valdman, B. 2001. Biomass monitoring by use of a continuous on-line optical sensor. Sensors and Acruators. 75(1/2), 24-28. van Holde, K.E.1985. Physical Biochemistry, Prentice-Hall, New Jersey, USA. Atkinson, B., and Mavituna, F. 1991. Biochemical Engineering and Biotechnology Handbook, Second edition, Stockton Press, New York.