Why do we need Hardware?
Current electrochemistry setup can cost over £1000(first figure below) as they include a counter and working electrodes made of materials such as gold as well as AgCl reference electrodes. While this kind of set-up has its own advantages such as great biocompatibility, outstanding conductivity and low reactivity with organic molecules, pure gold wires are too expensive to make it affordable and cost-effective. Instead, we opted for carbon electrodes which are up to 7 times cheaper(check setup in second figure below) than gold wire when comparing price per cm2 - the main geometric feature of interest in electrochemistry set-ups and 250 times cheaper when looking at the three electrodes required to stimulate gene expression.
Characteristics & Performance
To characterise our system a series of square wave voltammetry and cyclic voltammetry experiments were performed. For these experiments, a 90% content carbon working electrode of 2mm diameter was used, a 60% carbon counter electrode of 5mm diameter and a custom-made Ag/AgCl reference electrode was used . To drive the specific redox potential into our system amperometry experiments were performed. For these, the setup only differs in the use of 60% carbon working electrodes of 5mm diameter as opposed to 90% content carbon working electrode of 2mm diameter. A table for the technical specifications of these electrodes is found here:
|Electrode number||Electrode type||Length (mm)||Diameter(mm)||Specifications||Capacitance (S/m)|
|1||Working/Counter||100||5||60% Carbon - 40% Epoxy Resin||195 ± 11|
|2||Working||100||2||90% Carbon - 5% Clay - 5% Wax||796 ± 38|
This cheap 3-electrode setup was capable enough of stimulating gene expression as shown in the WetLab results section.
Comparison of price per setup
|Electrode type||Analysis Setup (£)||Driving Setup (£)||Literature Setup (£)||Total (£)|
Comparison of price per cm2
|Electrode type||Analysis Setup (£)||Driving Setup (£)||Literature Setup (£)||Total (£)|
Once we experimentally demonstrated how of gene expression could be locally induced with our affordable 3-electrode setup, the next goal was set to build a higher-end electrode array able to induce genetic expression locally at different regions in a lawn of cells.
The electrode array is printed onto one side of a Printed Circuit Board (PCB) in the form of 100 (10x10) circular working electrode pads, 4 counter rectangular pads, and 1 reference pad with the shape of a cross as it can be seen in figure XX ~(figure of pad configuration) . All the electronic components that allow controlling the logic of the array are placed on the other side of the PCB. We want each individual pad to select one of two working electrode potentials. The circuit can be broken down into several modules:
- Electrode cell: This corresponds to a memory storage unit, a set of two transistors and the electrode pad itself connected together. The storage unit is operated by a Data Flip-Flop and stores 1-bit (0 or 1), based on the data flip-flop output the transistors select one of the available Working potentials and outputs it into the pad.
- Power module: logic chips are powered through a voltage difference between a constant voltage supply (Vcc) and a ground(GND) supply. The logic HIGH (1) of components corresponds to a voltage of Vcc while the logic LOW (0) corresponds to a voltage equal to GND. We want logic LOW to correspond to 2.5V below the reference voltage of the electrochemical array, and logic HIGH to correspond to 2.5V above the reference voltage. this is to ensure that the Data Flip-Flops can force the FETs into either an ON or OFF state. Thus, we cannot use the real ground as logic GND. Instead, we run the logic (Arduino and PCB chips) from a non-grounded battery, in order to be able to shift logic levels up or down. This shifting is achieved by an op-amp, with its non-inverting input connected to the reference electrode. The op-amp is connected in negative feedback to a 2.5V voltage regulator. This means that it sources and sinks current as required to maintain its output (2.5V relative to logic GND) at the reference voltage, effectively centring the logic HIGH and LOW around the reference voltage. This ensures that the logic circuits can reliably switch the FETs.
- ROW-Selecting module: in order to program our electrode pads we need to sort through every single row and feed individual data into each pad. A demultiplexer selects electrode rows one at a time, telling each row in turn to read the data supplied to it.
- Data-feeder: The data, which is either a 1 or a 0 (HIGH or LOW) is fed into the individual cells of the selected row. This operation is accomplished by a serial-input-parallel-output (SIPO) shift register which outputs 10 bits at a time (one per column of the electrode array).
- Memory Unit - Data Flip-Flop: data flip-flops are simple devices that can be built with logic gates. DFF store one bit of information: HIGH or LOW. In our system DFFs are used to store the fate of the array cell and will give orders to the FETs on which potential to choose from.
- Selection Unit - MOSFETs: metal-oxide-semiconductor-field-effect transistor or MOSFET are in practice switched independently from any human mechanical input. MOSFETs behaviour is characterised by their terminals: Gate, Source and Drain and whatever electronic events are happening at these terminals. These minuscule devices respond to a certain voltage operating at the Gate terminal and will allow for the flow of electrons(i.e current) when certain conditions are met. whenever the voltage at the gate is greater than a certain value, electrons will flow, otherwise, they won't. In our system, MOSFETs are used to select for either of the potential coming from the potentiostat, either oxidising or reducing which will determine if the array cell will be an oxidising or a reducing cell.
- Row Selection Unit - Demultiplexer:demultiplexers are logic devices that allow selecting for one output line out of many available according to the logic input in the so-called select lines. In our circuit, one demultiplexer is used to select for a row of array cells which data is going to be fed into (from the shift register)
- Data feeder - Shift Register: shift registers are technically a row of DFF all connected together, upon a clock “order” a data flip-flop will share it’s “memory” with next data flip-flop hence whatever data is stored in the SR ( 1110) will be shifted to the right (0111). In our circuit, a serial to parallel SR is used to determine the value of the bits that are going to be fed into the DFF in the preselected row by the demultiplexer.
- Power control - opamp: operational amplifiers or op-amps are extremely useful devices that can be used for many applications. In our circuit, we use an op-amp for one of it’s most basic purposes: a voltage follower. A voltage follower “copies” the voltage from one terminal without drawing any current from it.
- Power control - voltage regulator: a voltage regulator transforms whatever voltage it has at its input to a specific voltage determined by its internal configuration. In our circuit, the voltage follower is an elegant and easy way to have the Arduino and the PCB running from one single battery. The Arduino runs on a 9V battery and the same voltage is then sent to the voltage regulator which then power the logic devices in the PCB.
- Capacitors: capacitors can be seen as tiny batteries, they are able to hold a charge and then release it. they have many applications of course but in our circuit, we use most of them as decoupling capacitors which damp the noisy oscillations coming from high-frequency noise sources.
Concept Design Justification
The concept design we present is composed of only 6 types of components. All of them are very well characterized, hence their performance is robust and predictable. Only one component needed to have particular specifications and that is the MOSFETs. In the case of the MOSFETs, we planned to connect the two drains together and use one input channel to interface with the gate of both the P-channel and N-channel FETs. This design runs the risk of having both FETs conducting at the same time, shorting out the circuit and damaging components, if badly designed. A more robust and typical solution to this problem would be to use a monolithic half-bridge control chip. This sort of chip takes an input signal and coordinates its two outputs so that the connected FETs will not connect simultaneously. However, their use would greatly increase the PCB surface area, energy consumption, heat generation and component cost. Therefore, we instead opted to use a pair of matched N-channel and P-channel FETs in a single package. The chosen components switch on at Vgs < -0.65V for the P-channel, and Vgs > +0.65V for the N-channel. Thus, for Vs = 0.5V (P-channel) and Vs = -0.5V (N-channel), there is a buffer region between -0.15V and +0.15V, within which both FETs will be off. Note that this is a simplification, as the current/voltage characteristic for the FETs is not a step response, but this region is sufficient to prevent damaging magnitudes of leakage current during switching. It should also be noted that this buffer region sets a hard limit on the voltages that the PCB can switch; if the differences between the two electrochemistry voltages exceed 1.3V, the circuit will short upon switching, which could damage the components. Our system is completely modular, hence if an array with more or fewer electrodes is desired, the number of electrode cells can be changed accordingly (with the respective components). Additionally, pad size can be modified to regulate the size of the region we wish to stimulate.
Design & Modelling
In order to ensure that our design would operate as intended, we constructed models of the MOSFET switching subcircuit using LTSpice. We chose a pair of well-matched MOSFETs, the P-channel Si9434BDY, and the N-channel Si5515_N, and performed a linear sweep of the gate voltage between -2.5V and +2.5V. As expected, as the gate voltage passed through 0V, a small amount of leakage current (1.8mA) passed through the two FETs due to an increase in their subthreshold conductance. However, this corresponds to a minimum effective resistance of 560Ω during switching, and this current is far too small to pose a threat to the components. For reference, the set of MOSFETs we have used can dissipate up to 490 mW, whilst the parasitic power from the leakage current is just 1.8 mW. Additionally, due to the rapid speed of switching of the FETs, and the correspondingly high slew rate of the D-type flip-flops, the duration of this quasi-short circuit is very short (a few microseconds in the worst case scenario), which further limits the energy wasted to ohmic heating. In conclusion, the model of the circuit confirms that it operates safely, and confirms that this simple control circuit is sufficient to switch these small voltages.
Board Architecture and Design Geometry
The PCB consist of 4 layers of copper:
- Front Copper: where the components are placed
- Second Copper: this layer is used as a ground layer, it can be connected through a 0-ohm resistor to either the reference electrode or to the logic GND terminal in order to minimise the noise as much as possible.
- Third Copper: internal connections go through this layer with the aim of reducing the size of the board and in order to avoid as much as practically possible populating the bottom copper layer with connections.
- Bottom copper: the electrode pads are on this side, which is the one in contact with the lawn of cells.
The printed circuit board (PCB) electrode array consists of:
- 100 working electrodes with a 5mm diameter with a total added area of 1963.5 mm^2 separated by 8mm from any other working electrode pad horizontally and vertically
- 4 rectangular counter electrode pads with dimensions 8.1175 by 77mm
- 1 reference electrode in the form a cross with dimensions 77 by 77 mm and width of 0.5mm
The geometry can be visualized here:
In order to confirm the desired operation of the switching circuit prior to construction of the array, we constructed a mock-up of the switching circuit. This comprised a D-type flip-flop, with two buttons to control its data and clock pins, with its output connected to the two driving FETs. We tested the control circuit with electrochemistry voltages of ±0.3V, ±0.4V, and ±0.5V. We placed a 100Ω resistor between the drain pins of the two FETs in order to be able to observe the current passing on an oscilloscope. As can be seen from the scope traces, at the time of switching, there is an oscillatory signal of magnitude 1V peak-to-peak and time period of 300ns. This signal seems to be heavily damped, only being discernable for one cycle. It can be seen that this signal is unbiased in either a positive or negative direction, meaning that there is no significant transfer of power in either direction during this switching period. This, combined with its extreme brevity, allow us to conclude that switching these voltages do not cause any significant energy loss. This conclusion can be additionally reinforced by contrasting the above traces with the oscilloscope traces taken with electrochemistry voltages of ±2.5V. From these traces, it can be seen that the voltage recorded during switching is heavily biased in the negative direction, which represents sustained current flow in one direction due to a short circuit during switching. Even in this case, however, when a short circuit is undeniably established, there was no noticeable heating of the FETs, or noticeable loss of switching performance, due to the nanosecond timescales of this short circuit. Indeed, even though ±2.5V exceeds the designed maximum switching voltage by a factor of 3.84V, the drain current through the FETs was 41mA, a safety factor of 130 below the component’s maximum drain current.
Price Breakdown - Bill of Materials (BOM)
The complete programmable electrode array costs £120 to make. An Arduino will also be necessary to control it and the app is free of any costs. In future versions, a microprocessor and the potentiostat could be embedded in the board in order to make the board an all-in-one device.
|Part number||Part description||Required units
||GBP/unit||Total price (GBP)||Used Supplier||Manufacturer||Manufacturer Part number|
|1||Printed Circuit Board (Gold-plated)||1||67.69||67.79||Beta Layout Inc||preferred||none|
|2||4-to-16 Demultiplexer||1||1.33||1.33||DigiKey||Nexperia USA Inc.||74HCT4514D,653|
|1||1.32||1.32||DigiKey||Nexperia USA Inc.||NPIC6C4894PW-Q100J|
|4||Single Output Data Flip Flop||100||0.103||10.3||Farnell Element 14||NEXPERIA||74AHC1G79GW,125|
|5||Dual N-P MOSFET||100||0.265||26.5||Farnell Element 14||NEXPERIA||PMCPB5530X|
|6||Operational Amplifier||1||0.967||0.97||Farnell Element 14||Texas Instruments||LMV651MG|
|7||2.5 Volts - Voltage Regulator||1||1.2||1.2||DigiKey||Texas Instruments||LP38690SD-2.5/NOPB|
|8||1 \μF Capacitor||2||0.058||0.12||Farnell Element 14||TDK||C1608X5R1C105K080AA|
|9||100 nF Capacitor||100||0.103||10.3||Farnell Element 14||MURATA||GCJ188R71H104KA12D|
|10||0 \Ω Resistor||1||0.0273||0.27||Farnell Element 14||Panasonic Electronic
|Part number||Part description||Link|
|1||Printed Circuit Board (Gold-plated)||Beta Layout|
|2||4-to-16 Demultiplexer||DigiKey - Shift Register|
|DigiKey - Demultiplexer|
|4||Single Output Data Flip Flop||Farnell - FlipFlop|
|5||Dual N-P MOSFET||Farnell - MOSFET|
|6||Operational Amplifier||Farnell - Op-Amp|
|7||2.5 Volts - Voltage Regulator||DigiKey - 2.5VR|
|8||1 \uF Capacitor||Farnell - C_1\u|
|9||100 nF Capacitor||Farnell - 100nF|
|10||0 \Ohm Resistor||Farnell - R_0\Ohm|
In Practice - the actual PCB
WE assembled the PCB and performed some preliminary measurements. Further Testing needs to be done. Exciting results are coming!