Wireless Passive Sensor for pH Monitoring inside a

Small Bioreactor

Sharmistha Bhadra1, Chris Dynowski

1, Warren Blunt

2, Mike McDonald

3, Douglas J. Thomson

1, Michael Freund

3,

Nazim Cicek2 and Greg E. Bridges

1

1Department of Electrical and Computer Engineering 2Department of Biosystems Engineering

3Department of Chemistry

University of Manitoba

Winnipeg, Canada

E-mail: bridges@ee.umanitoba.ca

Abstract— A wireless passive pH sensor for continuous remote

bioprocess monitoring is presented. The sensor is small enough to

fit inside a small bioreactor or test tube. It consists of a planar

spiral inductor connected in parallel to a varactor forming a LC

resonant circuit. A pH combination electrode made of an

iridium/iridium oxide sensing electrode and a silver/silver

chloride reference electrode, is connected in parallel to the varactor. As the medium pH changes, the voltage across the

electrodes varies, shifting the resonant frequency of the sensor. For in-fluid monitoring the sensor is hermetically sealed to

encapsulate, and reduce parasitic capacitive coupling and eddy

current loss. The resonant frequency of the sensor is tracked

remotely by an interrogator inductor inductively coupled to the

sensor. The sterilizable sensor was used for remote pH

monitoring of Yarrowia lipolytica fermentation in a shake flask

over 67 hours. Experiment shows that the medium pH can be

monitored with 2.46 MHz/pH sensitivity and maximum deviation

of 0.07 pH from a commercial pH probe measurement over a 6.5-5.26 pH range.

Keywords— Bioreactor, inductive coupling, non-invasive,

passive, pH, resonant frequency, wireless, Yarrowia lipolytica.

I. INTRODUCTION

Bioreactors are most commonly used for carrying out bioprocesses to produce many commodities and chemicals. During bioprocess, optimal cell growth depends on pH control and many cells produce acids as a metabolic by-product. Therefore, monitoring and regulating the pH of the medium is one of the key steps for successful bioreactor operation [1], [2]. Conventionally, sterilizable electrochemical pH probes are used for bioprocess monitoring. However they need wired connections for data exchange and are inherently invasive. Shake flasks and test tubes are regularly used in academia as well as in industry for selection and bioprocess development. The use of wired pH probes for multiple reactors (i.e. simultaneous monitoring of several shake flasks) requires a proportional increase of wiring, cost and complexity. Non-invasive sensors are an attractive alternative [3]-[6].

Although a large number of studies are being done on developing non-invasive pH sensors, only a small fraction can

be used inside bioreactors. Many sensors do not endure in the harsh bioprocess environment as the medium can permeate through and damage the sensor. Often the fluid medium culture is not well defined and interferes with sensor readings. Moreover, sensors used in bioprocess need to be sterilizable to avoid medium contamination and should not interfere with metabolism [1]. Non-invasive optical sensors based on absorbance or fluorescence from pH-sensitive dyes have been successfully used inside bioreactors. However, optical sensors typically suffer from a narrow operating range and drifting over time [7], [8].

In this paper we present a wireless passive pH sensor that can be integrated with and embedded inside a bioreactor. The sensor is robust and sterilizable. It is based on a passive LC resonator whose resonant frequency changes with the pH of the medium. An interrogator inductor is inductively coupled to the sensor and change in the sensor’s resonant frequency is detected by measuring the induced change in the impedance of the interrogator. The design of the sensor is simple and it is inexpensive to manufacture, making it a cost effective solution to non-invasive pH monitoring of a bioprocess.

II. SENSOR OPERATION

An equivalent circuit diagram of the wireless passive pH sensor is shown in Fig. 1, and includes the interrogation circuit/inductor coupled to it. In the remote sensor, a spiral inductor is connected in parallel with a voltage dependent capacitor (varactor) based voltage sensing circuit. A pH combination electrode is connected in parallel with the varactor and provides a biasing voltage to the varactor. The pH combination electrode in our sensor consists of an iridium/iridium oxide sensing electrode and a silver/silver chloride reference electrode. An Iridium/iridium oxide electrode has the advantages of easy preparation, small size, continuous detection ability as well as a fast and stable response in aqueous, non-aqueous, non-conductive, and even corrosive media. It can provide a linear response to pH with reference to a silver/silver chloride electrode over a 2-12 pH range with a low impedance. There is no requirement for pretreatment and has negligible interference of ions and

This research was funded by the Natural Sciences and Engineering

Research Council of Canada.

complexing agents. It has been used for pH measurements in technical media, such as fuels, food applications and in biological media [9], [10], [11], [12].

MRS

CVpH

VC

+

-

LS L1

Interrogator

coil

RpHR1

Sensor

coil

Sensing

circuitpH combination

electrode

Zin(f)

d

+

-

C1>>C

Low

pass

filter

Figure 1. Equivalent circuit diagram of the wireless passive pH sensor.

In the circuit shown in Fig. 1, LS and RS are the series inductance and resistance of the sensor inductor, respectively, and L1 and R1 are the series inductance and resistance of the interrogator inductor, respectively. M is the interrogator-sensor inductor coupling factor. C is the small signal junction capacitance of the voltage dependent capacitor (varactor diode) in the sensing circuit. In the reverse bias state, C is approximated by C(VC)=C0(1- VC/φ)-1/2, where C0 is the junction capacitance at zero bias, φ is the junction built in potential and VC is the bias voltage applied across the varactor. VpH and RpH are the potential difference and the cell resistance, respectively, developed at the pH combination electrode when in contact with a solution. In the circuit, a low pass filter was added so that the resonator is sensitive to low frequency variations in VpH. For a small interrogator source oscillation amplitude and small M, the approximation VC

VpH can be applied. As C1>>C(VC), the resonant frequency, f0, of the sensor can be approximated by,

)(2

10

pHSVCL

f

(1)

Therefore, pH of the contact solution, which is indicated by VpH, can be monitored by tracking the f0 of the sensor.

Referring to Fig. 1, near the resonant frequency, f0, the impedance, Zin, seen by the interrogator inductor is,

S

TinZ

M)f(fLjRZZ)f(Z

22

1112

2

(2)

where f is the interrogator source frequency and ZS≈RS+j2πfLS+(1/(j2πfC)) is the sensor series impedance. In our system the resonant frequency was obtained from the maximum of the real part of the impedance using a quadratic curve-fitting algorithm [13]. The impedance, Zin, in (2) consists of two components; Z1=R1+ j2πfL1, due to the self impedance of the interrogator inductor and ZT=(2πf)2M2/ZS, due to the sensor coupling. To remove the self impedance of the interrogator inductor a background subtraction, using the measured impedance of the interrogator inductor when the sensor was absent, was implemented prior to measuring the sensor response [14].

III. DESIGN AND FABRICATION

A. pH Combination Electrode

The iridium/iridium oxide (Ir/IrOx) sensing electrode was prepared by a direct oxidation method. Ir metal wire (0.5 mm in diameter, 99.8% purity, obtained from Alfa AESAR) of 10 mm in length was ultrasonically cleaned with 6M HCl solution followed with de-ionized water. The clean wire was then oxidized by bringing it to a temperature of 800ºC in an electric oven for 45 minutes after wetting its surface with 1M NaOH solution. The wetting and heating process was repeated six times until a blue-black coating was formed on the surface [10]. The electrode was immersed in boiling DI water for an hour and then in DI water at room temperature for 30 days to reduce aging effects. A small area of (about 2 mm) iridium oxide film at one end was scrapped off and connected to an insulated wire using silver epoxy. High temperature epoxy was applied over the connection area for electric insulation.

The silver/silver chloride (Ag/AgCl) reference electrode

was prepared from silver wire by electroplating method. 10

mm of Ag wire (1 mm in diameter, 99.8% purity, obtained

from Alfa AESAR) was polished using sandpaper. An

insulated wire was connected to a small area of the polished

Ag wire (about 2 mm in length) using silver epoxy. High

temperature epoxy was applied over the connection area for

electric insulation. A layer of AgCl was formed on the rest 8

mm of the polished Ag wire by applying +0.5 V for 50 s in a

0.1 M KCl solution. An immobilized electrolyte was freshly prepared by saturating 12 mL of THF with KCl at room

temperature and then adding 0.4 g of PVC. The Ag/AgCl wire

was dip-coated in the immobilized electrolyte solution and

dried in a house vacuum-evacuated desiccator for 48 hr. After

drying, the electrode was dip-coated with a protective nafion

layer to prevent the leakage of chloride ions. Following the

dip-coating the electrode was cured in a pump-evacuated oven

for 1 hour at 120°C, stored in a desiccator for seven days,

placed in DI water for 24 hours and then stored in a desiccator

until it was used [15].

B. Sensor and Interrogator Description

The sensor, as shown in Fig. 2, was constructed using the Ir/IrOx and Ag/AgCl electrodes described above. The sensor was designed to have a resonant frequency, f0, near 77 MHz. The sensor inductor and the voltage sensing circuit were fabricated on a 20 mm x 6 mm single sided FR4 printed circuit board (PCB) with surface mount capacitors and resistors. The inner and outer dimensions of the rectangular, four-turn planar spiral inductor were 6.5 mm x 1.2 mm and 12 mm x 5.5 mm, respectively, producing LS=0.13 μH and RS= 1.25 Ω at 77 MHz. The junction capacitance of the varactor (NXP BB202) used in the voltage sensing circuit, C, varied in the range of 35.04 pF - 22.95 pF for reverse bias voltages between 0 and 1V, respectively. For in-fluid monitoring the sensor was hermetically sealed in a polypropylene vial to encapsulate, and reduce parasitic capacitive coupling and eddy current loss. Locations where the electrodes came out of the vial were sealed properly with high temperature epoxy. The electrode resistance, RpH, for pH solutions in the range of 1.5-12 was approximately 3.16 MΩ. The Ag/AgCl and Ir/IrOx electrodes

were connected to the positive and negative terminals of the voltage sensing circuit, respectively.

The interrogator was fabricated on a 32 mm x 32 mm single sided FR4 printed circuit board. The inner and outer dimensions of the square, eight-turn planar spiral inductor were 3 mm x 3mm and 13 mm x 13 mm, respectively, producing LS=0.82 μH, RS= 2.2 Ω at 77 MHz and a self-resonant frequency, fres=97.3 MHz. The sensor’s resonant frequency was determined by measuring the real part of the impedance of the interrogator inductor, ReZin, when inductively coupled to the sensor. The interrogator inductor impedance was measured using an impedance analyzer (Agilent 4294A) with the voltage source level of the analyzer set to 25 mV.

Sensor coilSensing circuit

Polypropylene

vial

Reference

electrode

Sensing

electrode

Figure 2. Wireless passive sensor for bioprocess pH monitoring.

IV. EXPERIMENT AND RESULTS

As shown in Fig. 3, the sensor was embedded in a standard shake flask (bioreactor) containing the medium and carbon source. The medium was prepared from 10 g/L Yeast extract and 20 g/L peptone. Glycerol (13 mL of 50% glycerol solution to give a final concentration of 440 mM) was used as the carbon source. The total volume of the medium and carbon source was 200 mL. The shake flask and the sensor were both sterilized. A commercial pH probe and oxygen tubing were sterilized and inserted into the bioreactor to track the pH of the medium and supply oxygen, respectively. 40 mL of Yarrowia lipolytica cells were harvested during mid-exponential growth stage. This was centrifuged to separate the cells from the supernatant, and the liquid volume was reduced to 10 mL, concentrating the biomass in the inoculum by a factor of 4. Two mL of Yarrowia lipolytica cells were inoculated to the bioreactor filled earlier medium and carbon source for fermentation.

The medium was stirred at a constant speed of 100 rpm at 30°C. The resonant frequency, f0, of the sensor was monitored at half an hour intervals at a distance of 2 cm using the interrogator inductor and an impedance analyzer. A data logger (connected to the pH probe) stored the measured pH of medium. Fig. 4 shows the impedance frequency response of the sensor for two different medium pH values.

Variation of the sensor’s resonant frequency and medium pH (measured using the sensor and commercial pH-probe) over 67 hours is shown in Fig. 5. It can be seen that after a period of 2.5 hours, the pH monitored by the sensor was always in good

agreement with that measured by the commercial pH probe over the course of the fermentation period, with a maximum deviation of 0.07 pH. The initial difference in the wireless sensor result was due to stabilization of the reference electrode. The reference electrode was stored in desiccator and required 2.5 hours in the medium before its potential stabilized [15]. To test the repeatability of the sensor, base was added to the medium after 67 hours (when fermentation was completed) and pH of the medium was increased to 6.5 from 5.26 in two steps. Fig. 5 shows that the sensor produced repeatable results after the 67 hours test. This also indicates that sensor was not affected or damaged inside bioreactor by the medium.

Commercial pH

probe

Oxygen supply

tube

MediumInterrogator

inductor

Sensor attached to

the wall of the

shake flask

Impedance

analyzer

Figure 3. Yarrowia lipolytica fermentation experiment with the sensor

mounted inside the shake flask.

72 74 76 78 80 82 84

1

2

3

4

5

6

7

8

Resonant Frequency (MHz)

Re

ZT

Medium pH=6.37

Medium pH=5.26

Figure 4. Impedance frequency response for different medium pH values.

Fig. 6 shows the relationship between the medium pH (measured with the commercial pH probe) and the resonant frequency of the sensor excluding the data obtained during first 2.5 hours. A linear fit given by f0(MHz)=-2.46pH+91.983 over 6.5-5.26 pH range indicates a 2.46 MHz/pH sensitivity with a maximum deviation of 0.17 MHz (<0.07 pH) from linear fit. The effectiveness of the calibration equation determined from Fig. 6 depends on signal-to-noise ratio, ionic strength of the medium, cross sensitivity of the pH sensitive

electrode to other ions present in the medium and constant Δf offset due to the medium (this is minimized by encapsulation, but may still cause a small drift).

0 10 20 30 40 50 605.2

6.1

7

7.9

pH

0 10 20 30 40 50 6072.4

74.7

77

79.3

Time (hour)

Re

so

na

nt F

req

ue

ncy (

MH

z)

pH (probe)

pH (sensor)

Resonant frequency

Figure 5. Resonant frequency of the sensor and medium pH as measured

with a commercial pH ptobe and with the sensor using the linear calibration

curve in Fig. 6 over a period of 67 hours.

5.5 6 6.5

76

76.5

77

77.5

78

78.5

79

79.5

pH

Re

so

na

nt fr

eq

ue

ncy (

MH

z)

Resonant frequency

Linear fit

Figure 6. Resonant frequency of the sensor versus medium pH.

V. CONCLUSION

The wireless passive pH sensor based on a LC resonator is capable of real time and repeatable in-fluid pH measurement inside a small bioreactor. The sensor was successfully used in a shake flask set up to provide continuous remote measurement of the pH of Yarrowia lipolytica fermentation. The pH recorded by the sensor was in good agreement with values measured with a commercial pH probe with a maximum discrepancy of 0.07 pH over a 6.5-5.26 pH range. Although the accuracy of non-invasive optical pH sensors

(optical pH sensors have accuracy of 0.05 pH [16]) is better than the accuracy of the proposed sensor, narrow operating range and long-term drift due to photobleaching are the two concerns [7], [8]. Since the proposed sensor is based on eletrochemical probe, it operates over a wider dynamic range and does not have long-term drift problem.The robust and sterilizable sensor was not affected by the electrically lossy medium. Its size is suitable for pH monitoring inside a shake flask or test tube. The simple design of the sensor makes it a cost effective and non-invasive way to obtain reliable pH information from a bioprocess.

REFERENCES

[1] P. Harms, Y. Kostov and G. Rao, “Bioprocess monitoring,” Current Opinion in Biotechnology, vol. 13, no. 2, pp. 124-127, Apr. 2002.

[2] A. S. Jeevarajan, S. Vani, T. D. Taylor and M. M. Anderson, “Continuous pH monitoring in a perfused bioreactor system using an

optical pH sensor,” Biotechnology and Bioengineering, vol. 78, no. 4, pp. 467-472, May 2002.

[3] S. Vuppu, Y. Kostov and G. Rao, “Economical wireless optical ratiometric pH sensor,” Meas. Sci. Technol., vol. 20, pp. 045202(7pp),

Feb. 2009. [4] C. Komives and R. S. Parker, “Bioreactor state estimation and control,”

Current Opinion in Biotechnology, vol. 14, no. 5, pp. 468-474, Oct. 2003.

[5] A. Vasala, J. Panula, M. Bollók, L . Illmann, C. Hálsig and P. Neubauer, “A new wireless system for decentralized measurement of physiological

parameters from shake flask,” Microbial Cell Factories, vol. 5, no. 8, pp. 1-6, Feb. 2006.

[6] S. Kumar, C. Wittmann and E. Heinzle, “Minibioreactors,” Biotechnology Letters, vol. 26, no. 1, pp. 1-10, Jan. 2004.

[7] W. -L Tsai, J. L. Autsen, J. Ma, T. Hudson and J. Luo, “Noninvasive optical sensor technology in shake flasks for mammalian cell cultures,”

BioProcess International, vol. 10, no. 1, pp. 50–56, Jan. 2012. [8] H. R. Kermis, Y. Kostov, P. Harms and G. Rao, “Dual excitation

ratiometric fluorescent pH sensor for noninvasive bioprocess monitoring: development and application,” Biotechnology Progress, vol.

18, no. 5, pp. 1047-1053, Sept. 2002 . [9] W. D. Huang, S. Deb, Y. S. Seo, S. Rao, M. Chiao and J. C. Chiao, “A

passive radio-frequency pH-sensing tag for wireless food-quality monitoring,” IEEE Sensors J., vol. 12, no. 3, pp. 487-495, Mar. 2012.

[10] G. Papeschi, S. Bordi, M. Carlagrave, L. Criscione and F. Ledda, “An iridium-iridium oxide electrode for in vivo monitoring of blood pH

changes,” Journal of Medical Engineering & Technology, vol. 5, no. 2, pp. 86-88, Mar. 1981.

[11] P. Kurzweil, “Metal oxides and ion-exchanging surfaces as pH sensors in liquids: state-of-the-art and outlook,” Sensors, vol. 9, pp. 4955-4985,

June 2009. [12] S. Yao, M. Wang and M. Madou, “A pH electrode based on melt-

oxidized iridium oxide,” Journal of the electrochemical society, vol. 148, no. 4, pp. 29-36, 2001.

[13] M. P. Robinson and J. Clegg, “Improved determination of Q-factor and resonant frequency by a quadratic curve fitting method,” IEEE

Transaction on Electromagnetic Compatibility, vol. 47, no. 2, pp. 399-402, May 2005.

[14] J. B. Ong, Z. You, J. Mills-Beale, E. L. Tan, B. D. Pereles and K. G. Ong, “A wireless, passive embedded sensor for real time monitoring of

water content in civil engineering materials,” IEEE Sensors J., vol. 8, issue 12, pp. 2053-2058, Dec. 2008.

[15] M. A. Nolan, S. H. Tan and S. P. Kounaves, “Fabrication and characterization of a solid state reference electrode for electroanalysis of

natural waters with ultramicroelectrodes,” Anal. Chem., vol. 69, pp. 1244-1247, Mar. 1997.

[16] Non-Invasive pH Sensors [Online]. Available: http://www.presens.de/fileadmin/user_upload/products/Sensor_Probes/N

on-invasive_pH_Sensors/110715_SP-NonInvpH-11-01_w.pdf.