Bacterial cultures, especially in bulk quantities must be closely monitored to grow in optimal conditions. These conditions include temperature, pH and oxygenation.
Previous projects have effectively tackled the first two components of this system, but the dissolved oxygen sensing has remained an challenge not only for the Biomaker challenge-inspired bioreactor but in similar initiatives as well.
In this project, we started to build a low-cost oxygen sensor to measure dissolved oxygen in bacterial cultures. We developed a three electrode system device. Gold wire is used as a cathode and records current from the oxidation process as the analog input of the sensor. A silver / silver chloride electrode (Ag/AgCl) commonly used in electrocardiogram (ECG) is used as a reference electrode, while graphite, which could be sourced from a pencil lead, is used as an auxiliary electrode to further stabilize the system. The polarization of the electrode will be controlled by a potentiostat. We have identified components that will enable us to build a robust and reliable potentiostat.
The overall aim is to have a cheap, open source device able to record the dissolved oxygen in a batch E. coli culture in a bioreactor.
Bacteria fermentation is the cornerstone of industrial biotechnology. The growth of large quantities of bacteria in a controlled environment enables not only the reliable production of proteins or compounds but also the production of large-scale cell-free systems which are being used for prototyping genetic circuits and rapidly advance molecular biology research. In order to achieve the best metabolic performances, bacterial growth must be constantly monitored through parameters such pH, temperature and dissolved oxygen through the use of appropriate sensors.
Previous published projects (Microbial bioreactor) have tackled pH, temperature and OD monitoring. In this project we will design a low-cost sensor for the monitoring of dissolved oxygen. Existing commercial solutions are prohibitively expensive for low-resourced laboratories and do not allow easy customization and maintenance. Moreover, to the best of our knowledge, none of the available full-documented open source low-cost bioreactor systems provide a documentation to build a sensor for the monitoring of dissolved oxygen  a crucial parameter that allows to achieve high biomass yields when combined with the addition of feed.
Building a oxygen sensor requires skills that are often beyond biology training and this might be one of the reasons that explains the current lack of documentation to build low-cost, open source and customizable sensors to monitor dissolved oxygen in bacterial cultures.
We will use the workhorse of molecular biology Escherichia coli. In this project, the oxygen dissolved in a E. coli culture will be monitored with a low-cost oxygen sensor. During building trials, the monitoring will be done in baffled flasks, with the final goal to test the system in the open source bioreactor.
- Sterile or autoclavable parts
- Easily sourced: we will use commonly and globally available components wherever possible and seek to avoid vendor lock in.
- All designs will be shared in open file formats that can be edited with free software and under maximally permissive open hardware/software licenses
- Documentation will be sufficient not only to build the device but also to understand the functionality of each part.
- Protocols will be available when needed to calibrate and use the device effectively.
The first objective of the project is to create the electrode system. The oxygen concentration in the culture medium will be measured electrochemically and continuously for at least 24 hours. Gold is the choice of working electrode that is a stable and inert material to electrochemically reduce oxygen. Gold wire encased in epoxy will be made the working electrode. It will have a fine opening at one end exposing the cross-section of the gold wire to the electrolyte. The opening will be protected by a oxygen permeable membrane resistant to cell attachment and protein aggregation. PTFE and cellulose membrane (dialysis membrane) will be investigated for their applicability to the current set-up. PTFE is cheap, hydrophobic and oxygen permeable while dialysis membrane have different pore sizes to give different diffusivity for different sizes of molecules. The selectively permeable membrane will be held at the opening with an O-ring encapsulating a small volume of electrolyte. A silver rod will be the reference electrode placed in proximity to the working electrode.
The second objective of the project is to develop a cheap small-scaled potentiostat. A potentiostat is highly important for controlling electrode polarization and measuring the current output which correlates to dissolved oxygen concentration. Using an op-amp and few electronics could create a simplified version of potentiostat (fig. 3). Oxygen concentration can thus be measured and calculated in real-time by Arduino which can execute the subsequent actions (e.g. introduce additional feeding). A touchscreen user-friendly interface will be set up and programmed for easy control of functions (e.g. calibration and oxygen measurement) and displaying oxygen concentration in real-time.
Oxygen sensing electrode - Gold electrode
Oxygen concentration can be sensed in multiple methods including optically, chemically and electrochemically. Herein, we used electrochemical oxygen sensors for its sensitivity, cost-efficiency and durability. We selected gold as the material of the working electrode. Polarized by the potentiostat (we are currently using a commercially available potentiostat), the electrode could reduce dissolved oxygen molecules into hydrogen peroxide or water molecules via oxygen reduction reaction. The measured current correlates to the oxygen concentration existing in the solution, giving a relatively accurate quantitative read-out of the amount of dissolved oxygen.
The area of the working electrode in contact with the solution to be monitored must be small compared to the reference electrode. Therefore, the gold wire needs to be encased in epoxy resin, leaving only the tip of the wire exposed to the solution. In order to fabricate the gold electrode we first intertwine the gold wire to a copper wire, fixing the contact using a conductive silver-epoxy resin (Fig. 1.1). We let the silver epoxy dry over night and we proceed with the encasing the electrode in clear epoxy. To do so we needed a mold, and we wanted to create a flexible mold that we could reuse over time. We used liquid latex poured in a petri dish (any small container will do) and use the inner plastic tube of an exhaust pen (the tube that encased the ink) to form the cylindrical shape on the surface of the liquid mold (Fig. 1.2). We let the liquid latex to dry for two days. The plastic tube formed a mold suitable to encase the working electrode. We filled the cylindrical shape with clear amber epoxy and gently push the exposed copper and gold wire into the epoxy resin (Fig. 1.3). We let the epoxy sit overnight. It is of foremost importance that the epoxy used in this stage does not form bubbles. We cut the tip of the encased electrode with a scissor and we polished the end with fine aluminium sand paper. The gold end of the wire is now clearly visible (Fig. 1.4).
Having a gold electrode fabricated, we characterized the electrochemical properties for oxygen sensing. We performed cyclic voltammetry in air and degassed environment and identified the presence of oxygen reduction peak in the system at -0.65V (Fig. 2). This potential was chosen for future polarizing potential for oxygen measurements. We measured the resultant current under different oxygen concentrations and reproduced the correlation between the two as known in literature. This allows us to use the established system as an oxygen quantitative sensor.
Protein fouling has long been an obstacle for oxygen sensors due to the reduction in sensing capability over time. We investigated the electrode stability in one of the most commonly used culture medium, LB broth (Fig. 3). We measured the oxygen concentration of sterile LB broth without bacteria over 24 hours under air. No apparent change in measured oxygen concentration was observed. This suggests that protein fouling was not severe in this culture medium. Nonetheless, we investigated the use of Teflon membrane and cigarette paper as the protective membrane of gold electrode. Significant reductions in current were observed in both cases in which current under Teflon membrane is beyond detection sensitivity of the system. This is reasoned by the reduction of diffusion of electrolyte and oxygen as the membrane acted as a barrier.
We first investigated the oxygen sensing capability of the electrode system by introducing reduction in ambient oxygen concentration (Fig. 4). The electrode was first stabilized for 30 minutes and a nitrogen gas was introduced to the ambient environment. Shortly, the measured current reduced in logarithmic shape and reached stability after 3 hours. This was evident that the electrode was capable in responding to the change in dissolved oxygen concentration in the electrolyte. We further addressed the effect of convection on measured current, which affects the oxygen reading (Fig. 5). We demonstrated that under convection, the measured current is significantly increased despite ambient oxygen concentration is kept constant. Convection is commonly used in bioreactors to increase oxygen availability to the bacteria due to slow diffusion of oxygen. This fluid motion would affect the current reading and affect the calculated oxygen concentration if not taken into account.
The criteria of a good reference electrode depends on its stability, which can be quantified by the potential drift over time and upon polarization. Sourcing for the cheapest available reference electrode, we decided to choose silver / silver chloride as our pseudo-reference electrodes for their stability over time and user-friendliness. They are often used in ECG probes which has become our source of interest. We narrowed our choices to Ag/AgCl ECG probes from 4 different suppliers, offering different designs, costs and functionalities. The cheapest electrode from Becamed has a small surface area and relatively low quality control such that it was not considered for further experiments. The remaining 3 suppliers, Funspark, 3M and Tianrun Sunshine, have larger surface areas with higher quality control and were further characterized (Fig. 6).
To be able to use the reference ECG electrode in a solution we connected the tip of the ECG to a copper wire using silver epoxy. We then encased all the conductive material but the Ag/AgCl into clear epoxy (Fig. 7).
All Ag/AgCl electrodes arrived with a hydrogel coating covering the entire electrode surface. Electrode potential stability were studied in phosphate buffered electrolytes both with and without the hydrogel coating (Fig. 8-10). With coating, electrodes showed a slow and gradual drift in potential against a laboratory-tested reference electrode (Fig. 8,9). This might be due to the gradual equilibration of electrolyte across the hydrogel and solution, resulting in a change in charged particles on the surface of the electrodes. The significance in potential drifting makes the reference electrodes with hydrogel on not ideal for the current application. When the hydrogel coatings were removed, all electrodes but Tianrun Sunshine showed improved stability for long periods of time (Fig. 10). Therefore, we decided to use Funspark and 3M as our reference electrodes.
We also investigated the feasibility of using a more compact 2 electrode system or a more stable 3 electrode system for this set up. Under the 2 electrode system, the reference electrodes broke down and showed potential drifts. This left us with no choice but to use a 3 electrode system for future measurements. The third electrode selected is graphite which can be easily sourced from a graphite pencil lead (Fig. 11).
Typically the cyclic voltammetry to measure the output of the oxygen sensor would be performed with a lab potentiostat, costing several thousand pounds. We need to do this accurately for much lower cost. Fortunately there are now several integrated circuit “analog front ends” from different manufacturers, that purport to integrate much of the electronics for typical potentiostat uses. In particular the Analog Devices AD5940 or AD5941 (same chip, different packaging) announced in July 2019 is very attractive. As well as the analog signal conditioning it includes features such as a 12-bit digital to analog converter to generate a bias voltage (0.2 V to 2.4 V), a programmable gain amplifier, and a 16-bit 800000 samples per second analog to digital converter.
At around £7 for the chip (in small quantities) it is probably cheaper and more reliable than selecting and combining several alternate precision parts in a circuit. It has been decided to acquire an evaluation system for the circuit, to test with the oxygen sensor. Also to evaluate the processing power required for the system - the evaluation system includes an ARM Cortex-M3 controller board. More details of the chip (135 page datasheet) are available .
Final design of a compact electrode: we are working on designing a compact version of the sensor. Ideally, the three electrodes will be housed in the same epoxy housing. Depending on the final design, the electrode might be partially autoclavable.
Coating studies: to avoid the fouling of organic substances on the surface of the electrodes we would need to coat the electrodes with a gas permeable membrane. We have been testing a Teflon membrane that is used in similar devices . However, the fabrication process needs to be further improved to achieve an optimal coating.
Oxygen monitoring in bacterial culture
Potentiostat: presuming the potentiostat chip AD5941 works well with oxygen sensor, we will design a circuit board with chip, connectors for oxygen sensor, and a microcontroller of sufficient power to process the data generated. Most likely a typical Arduino Uno is not powerful enough.
Software: a software will be written in order to a) calibrate the oxygen probe b) monitor the oxygen concentration over time and c) create a feedback loop with a feeding solution for the bacterial culture.
Optical sensor for oxygen monitoring: if the electrodes system described above won’t satisfy the requirements to have a low-cost sensor, we will explore other options. One of them is building an optical oxygen sensor. However, if this could be a viable option needs still to be determined.
Documentation: progress of trial and errors will be updated on a rolling basis. Full documentation of the final sensor and potentiostat design and software will be available as soon as defined.
Other less capable integrated circuits with aspects of a potentiostat’s analog signal processing are the Texas Instruments LMP91000 and the New Japan Radio NJU9101.