HARDWARE

To ensure SPYKE can be used properly, the GHB sensor has to be integrated in a piece of hardware. Here, we present the design of a dedicated glass containing the GHB sensor. We were able to build the prototype and assess its usability and functionality.

Introduction

With SPYKE, we aim to create a GHB biosensor that can prevent GHB spiking by measuring a capacitance change. To integrate our sensor optimally into the nightlife scene, we designed a drinking glass in which our bioelectric sensor can be incorporated. The glass was developed to prevent the occurence of drink spiking by continuously measuring the GHB concentration in beverages. A video of our 3D-printed prototype can be seen in Figure 1 .

Figure 1. A video explaining the final design of our prototype.

The device detects GHB by continuously measuring the capacitance of electrodes. The DNA-immobilized electrodes are incubated with the allosteric transcription factor BlcR which binds to the DNA and is responsive to GHB and the GHB-analog SSA. When GHB is added to the solution, BlcR unbinds from the DNA, resulting in a capacitance change. Figure 2 shows the mechanism used to detect GHB.

Figure 2. Simplified overview of the mechanism used to detect GHB.

Inside the device electronic circuitry is present which continuously measures the capacitance of the electrodes. When the capacitance changes due to the addition of GHB a light turns on warning the user. After designing our prototype, we 3D-printed and assembled the parts into a final sensor. A figure of the 3D-printed prototype can be seen in Figure 3 . We ran experiments and simulations to test and improve the sensor's performance. We let a possible user and a bartender review the prototype and proposed future design changes to improve its performance. We share the full building guide which can be used to recreate the hardware device for around 60 euros.

Figure 3. An image of the final 3D printed and assembled hardware device.

Integrated design

Interaction with stakeholders were essential to create an impactful device to prevent GHB spiking. Therefore, we conducted many interviews, surveys, and email interactions through which we obtained valuable insight from stakeholders and experts. This knowledge was subsequently used to shape the design of our hardware. A full overview of the stakeholder interactions is described on the integrated human practices page.

Morphological chart

To visualize the design choices resulting from the stakeholder interactions we decided to use a morphological chart. A morphological chart is a way of describing possible ways of achieving functionality. The first column of the chart describes the required functionality. The second, third, and fourth columns describe possibilities for achieving the functionality with the preferred one highlighted. The final fifth column explains why we choose this possibility. The morphological chart is shown below.

The morphological chart is shown in Table 1.

Table 1. A morphological chart showing the design choices made influenced by stakeholder input. The chosen design is highlighted
Functional requirement Concept 1 Concept 2 Concept 3 Explanation
GHB detection mechanism Fusion protein Capacitance measurement - We use a capacitance measurement to detect GHB inspired by Frans Widdershoven as it can be combined with a clear cutoff value. We learned from Marc Ostemeijer that the creation of a fusion protein will take a long time. Also, a fusion protein will result in leakage as a low endogenous GHB concentration in drinks will cause continuous activity [1] .
Sensor placement Glass Stick Tooth We incorporate the biosensor in a glass as the survey results clearly showed this was the preferred option for the users. The respondent’s main reasons were that the stick isn’t passive and the tooth is invasive and gives a signal after the GHB was ingested.
Output signal LED Bluetooth - EHBDD, clubs, and the police recommended using a LED as an output. The reason for this was that the Bluetooth signal could be easier to miss for the user and the clubs. Also, it can be hard for the club to find the glass with the positive signal.
Number of capacitors One Two - We chose the more expensive two-capacitor system to reduce the environmental influences mentioned by Jeroen Bastemeijer.
Signal processing Arduino UTI Rasberry Pi We choose Arduino to do signal processing. As advised by Jeroen Bastemeijer. the Arduino is capable of doing all the tasks necessary while being cheaper than the more complex UTI and Rasberry Pi.
The number of 3D printed parts (for more information see the section 3D printed glass) Three parts including one solid cup with a nut and a bolt system to place the electrodes. Four parts including a solid cup, a separate compartment for the Arduino, and a nut and bolt to place the electrodes. Five parts including compartments for the beverage, the electrodes, the Arduino, and a nut and bolt to place the electrodes Together with Nemo Andrea we designed a five-part system to allow for easy replacements of parts to decrease the waste.
3D printer Filament PLA ABS - PLA and ABS were the two available 3D printer filaments. We choose PLA due to its increased strenght and biodegradability. PLA is less heat resistant than ABS, but our device will not be subjected to heat [2] .
Filter Ultrafiltration discs Dialysis membrane - Ultrafiltration discs are meant to be used under pressure for a short time [3] as opposed to dialysis membranes which are meant to be used without pressure for a long time [4] . To ensure the filter works safely throughout the night we opted for using dialysis membranes.
Waterproofing O-rings Glue Clips To make the division between the various 3D printed parts watertight we used O-rings [5] . Glue would not be a good fit as it would limit the replacement of parts. Special clips would have to be made for our exact design making it harder to recreate.
User manual QR code on the glass with link to the manual. Website Paper manual Inspired by Lars van Driel’s advice we chose to put a QR code with a link to the user manual on the glass. This is a more user-friendly way of presenting the user manual than a website or paper manual.

Prototype

The design choices selected in the integrated design were all combined into our first prototype. The device consists of three parts: the electronic readout circuit, a 3D-printed plastic glass, and a dialysis membrane separating the protein solution from the beverage.

Electronic circuit

The electronic circuit present in SPYKE measures the capacitance value of an electrode and turns on a light when the capacitance change exceeds a certain threshold. To achieve this, we used an Arduino Nano board [6] in our prototype to measure the capacitance. We also used the open-source software LTspice [7] to model the electronic circuits to analyze the influence of changes to the circuit.

Arduino

For the prototype we use an Arduino nano board on advice from Jeroen Bastemeijer. The board is capable of continuously producing and measuring voltages while being connected to a power source. The Arduino can be connected to a laptop and by using the Arduino IDE software, the microcontroller can be programmed. With a simple code, the system continuously measures the capacitance, takes the moving average to stabilize the signal [8] , and when a large change occurs it turns on a LED light. To power the Arduino Nano a simple 9V battery was used. [9]

Capacitance to voltage converter

The Arduino can only measure voltages, not capacitance values. A capacitance-to-voltage converter (CVC) was used for the prototype to convert a capacitance change into a voltage change. The Arduino can measure a voltage signal ranging from 0 to 5 V and maps the measured voltage between 0-1023 relative Arduino units. [10] We constructed an electronic circuit consisting of only the CVC of the universal transducer interface (UTI). We used LTspice simulations and experiments to verify the effectiveness of our circuit. We started with a circuit containing one capacitor, but to minimize environmental influences we upgraded to a dual-capacitor system. Gerard Meijer warned us the capacitor could be sensitive to leakage, a phenomenon where current flows through the capacitor when a DC voltage is applied. [11] We measured the leakage and simulated the consequences resulting from it.


The first system we developed continuously runs a square wave generated by the Arduino through one capacitor. The Arduino measures the output of the CVC. A schematic of the system is displayed in the Figure 4a . We ran an LTspice simulation varying the capacitance of the electrode and measuring the resulting output voltage Figure 4b . The chosen capacitances are in the range of the capacitances of the electrodes.

Figure 4. (a) A sketch of our one capacitor electrical circuit. Vdrive and Vo-CVC represent the voltage signal created by the Arduino and the output of the CVC respectively. Cx, Cp, and R3 represent respectively the electrode, a reference capacitor, and a reference resistor. (b)The results of an LT spice simulation of the output voltage from the CVC for different capacitor values of the electrode (pink line C=0.7 pF, black line C=0.8 pF, blue line 0.9 pF)

The LT spice simulations show the system is capable of measuring capacitance changes. We also implemented the circuit using Arduino, the circuit can be seen in the Figure 5. This circuit just like the single capacitor circuit, generates the square waves needed to power the CVC and measures the peak voltage values shown in Figure 4b

Figure 5. A schematic overview of the Arduino circuit that was used to measure the capacitance change.

The measurement setup was validated using capacitors with known values. With these capacitors, the output of the Arduino was measured. As shown in Figure 6, the Arduino output shows the expected linear increase resulting from capacitance changes and can thus be used to accurately measure changes in capacitance.

Figure 6. Validation experiments for the dual capacitor CVC. The difference in capacitance between the capacitors is plotted against the Arduino relative output.


Jeroen Bastemeijer told us there are problems with the one capacitor system as environmental changes can affect the capacitance [12] . Since these capacitance changes can be in the same range as the capacitance changes caused by the addition of GHB, this may cause a false positive signal. In Dr. Ali Heidari’s thesis [12] , another kind of capacitance-to-voltage-converter was mentioned which could overcome these problems. This CVC uses two almost identical capacitors and measures the difference between them. When the capacitance changes due to environmental changes, the difference between the capacitors stays relatively stable. This way, the influence of the environment is minimized. The difference is measured by sending two inverse square waves through the capacitors. The signals are then added afterward. The Arduino measures the output of the system.

In our project, our goal was to measure the change in capacitance caused by the unbinding of BlcR that is initiated by the addition of GHB. We used two electrodes, the first electrode contains immobilized DNA with a binding sequence for BlcR. The second electrode is identical, except the immobilized DNA doesn’t contain the BlcR binding site. The capacitance of the second electrode is therefore not responsive to GHB since there is no binding and unbinding of BlcR. Both of the electrodes respond similarly to external influences except for the addition of GHB. Measuring the difference between the two electrodes minimizes the environmental influences while still measuring for GHB. See Figure 7& 8 .

Read-out
Figure 7. Schematic overview of the factors that influence the capacitance of the electrodes.
Read-out
Figure 8. Schematic overview of the factors that influence the biosensor read-out in a two-capacitor configuration. The electrode on the left contains DNA with the BlcR binding sequence. The electrode on the right contains scrambled DNA. Only BlcR (un)binding influences the system significantly, which increases the signal-to-noise ratio.

A schematic of this dual capacitor system is displayed in the Figure 9a . We ran an LTspice simulation varying the capacitance of one of the electrodes and measuring the resulting output voltage Figure 9b .

Figure 9. (a) A LT spice sketch of our dual capacitor electrical circuit and the relevant signals. Vdrive and -Vdrive represent the voltage signal created by the Arduino and Vo-CVC output of the CVC. Cx, C1, Cp, and R3 represent the electrode, a reference electrode a reference capacitor, and a reference resistor respectively. (b) The results of an LT spice simulation of the output voltage from the CVC for different capacitor values of the electrode (purple line C= 0.5 pF, light blue line C=0.6 pF, blue line C=0.7 pF, black line C=0.8 pF, pink line 0.9 pF)

The LT spice simulations show the system is capable of measuring the differences between the electrodes. As can be seen in Figure 9b , the CVC outputs zero voltage when the capacitances of the electrode and the reference electrode are identical. We also implemented the circuit using Arduino as can be seen in the Figure 10 . This circuit just like the single capacitor circuit, generates the square waves needed to power the CVC and measures the peak voltage values shown in Figure 9b .

Figure 10. A schematic overview of the Arduino circuit that was used to measure the capacitance change.

The measurement setup was validated using capacitors with known values. We measured the output from the Arduino resulting from the difference between capacitors. The results are shown in Figure 11 .

Figure 11. Validation experiments for the dual capacitor CVC.

As can be seen from the LTspice simulations and the Arduino measurements the CVC accurately converts the difference between capacitance values to a voltage value. The output voltage linearly depends on the difference of capacitance. The formula of the trendline was used in future experiments to convert the relative output to capacitance values.


Leakage of the electrodes is a phenomenon where current flows through the capacitor when a DC voltage is applied and can influence the output [11] . To see if this would still result in problems for the prototype with the two capacitor measurements, we used LTspice to simulate the influence of leakage when measuring capacitance changes. We used a resistor in parallel with the electrodes to simulate the leakage. We used a resistance of 140 MΩ as was measured during the vector impedance measurements. The schematic used can be seen in Figure 12a . We ran the model of the CVC over a range of capacitance and resistance differences between the electrodes. The results can be seen in Figure 12b .

Figure 12. (a) A LT spice sketch of our dual capacitor electrical circuit with leaky capacitors integrated and the relevant signals. Vdrive and -Vdrive represent the voltage signal created by the Arduino and Vo-CVC output of the CVC. Cx, C1, Cp, and R3 respectively represent the electrode, a reference electrode, a reference capacitor, and a reference resistor respectively. Rref and Rx represent the leakage of the electrodes. (b) The results of an LT spice simulation of the output voltage from the CVC for different capacitor values of the electrode (blue lines C= 0.7 pF, purple lines C=0.8 pF, pink lines C=0.9 pF). The leakage resistance was varied for all capacitance values (130 MΩ, 140 MΩ, 150 MΩ).

As can be seen from the results of the LTspice simulation, the dual-electrode system would work well, even with a varying leakage between the electrodes as capacitance differences can clearly be measured. For this reason, the dual-capacitor system is incorporated into the prototype.



After analysis of the single and dual-capacitor system we concluded the dual-capacitor system contributes to a more reliable prototype. This is because, it reduces environental influences and decreases the problems caused by leakage of the electrodes. For these reasons we incorporated the dual-capacitor system into our prototype.

Determination of threshold value

To determine at what capacitance change the LED light should turn on to alarm the user, we first measured the change of capacitance of a single electrode induced by the addition of SSA. As we saw during the Arduino measurements, a change of around 6 relative values was caused by the addition of SSA. The maximum noise observed was less than 0.5 relative values. As a midway, we chose 3 relative values as the cutoff value for the light to turn. Using the trendline shown in Figure 11 we calculated this corresponds to a capacitance change of 1.6*10-2 pF.

Button

As we found out, from the various capacitance measurements, there was a large variance in capacitance and capacitance changes between the electrodes. This causes the Arduino to measure a difference in capacitance between the two electrodes without the addition of SSA. This offset capacitance difference will result in problems when setting the threshold. To ensure the prototype functions well, we added a button that removed this offset by baselining the capacitance difference to zero when pressed. This ensures that only the difference in capacitance resulting from GHB addition is measured. With the addition of the button the electronic circuit was completed and shown in Figure 13 .

Figure 13.The read-out circuit used to measure the capacitance changes. The button and LED light are located outside the glass.

3D printed glass

The previously described electronic circuit is incorporated in a 3D printed glass which can be seen in Figure 1. The glass is made up of five separate 3D printed parts which can be assembled into one glass. Using this five-layer system parts can easily be replaced making the device more sustainable as less waste is created. The parts that have to be regularly replaced are the electrodes, protein solution, and filter [4] .

  1. The beverage compartment: The part of the glass which will contain the beverage. It has a hole at the bottom where the filter is placed. It can be screwed onto the electrode compartment together with an O-ring to make the system watertight.

    2. Lunch lecture Hooke
    Figure 14. An animation and picture of the beverage compartment.

  2. The electrode compartment: The middle part of the glass is placed underneath the beverage compartment and separated by the filter. This compartment will contain the protein solution and has two holes in the bottom leading to the Arduino compartment.

    3. Lunch lecture Hooke
    Figure 15. An animation and picture of the electrode compartment.

  3. The Arduino compartment: The bottom part of the glass containing the Arduino and the electronic read-out circuit. It has a hole in the side to allow for a cable to connect the Arduino to a laptop. It has two protrusions that can be used to click it onto the electrode compartment.

    4. Lunch lecture Hooke
    Figure 16. An animation and picture of the Arduino compartment.

  4. The bolt: A bolt into which the electrode is placed. Two bolts will go through the holes in the electrode compartment to seal it from the Arduino compartment.

    5. Lunch lecture Hooke
    Figure 17. An animation and picture of the bolt.

  5. The nut: A nut which can be screwed on the bolts. Two nuts can be used together with two O-rings to tighten the bolts and make the system watertight.

    6. Lunch lecture Hooke
    Figure 18. An animation and picture of the nut.

Filter

The last component of the prototype is the filter. The filter is located between the beverage and the electrode compartment and its function is to separate the two. The filter should allow GHB to diffuse from the beverage to the electrodes while retaining the protein in the electrode department. We chose to use a dialysis membrane as a filter as they are designed to retain proteins and allow the diffusion of small molecules over a long period of time [4] . We conducted two experiments to test if the filter worked as intended. We first tested if the connection of the filter was watertight to ensure only diffusion through the filter occured. Afterward, we tested if the filter indeed allowed small molecules such as GHB to diffuse while retaining proteins.


It is essential that the connection between the beverage and the electrode compartment is watertight, to ensure that the protein doesn’t leak into the beverage. Nemo Andrea suggested using O-rings to decrease the leakage of liquid. We conducted an experiment to examine if our 3D-printed prototype is indeed watertight. A watertight, plastic, flexible layer on top of an O-ring was attached between the electrode and beverage compartment and tightened by screwing the compartments on each other. First, the cup was filled with a small layer of water and shaken on a plate for 30 minutes. Afterward, the cup was examined for any leakage from the top to the bottom. Then, the cup was fully filled with water and the process was repeated. The results are shown in Table 2.

Table 2. Overview of the results of the watertight testing
The volume of water Leakage
50ml 400ml
Yes No

We observed no leakage with a small volume of water, but when the cup was fully filled the cup almost immediately started leaking. These results indicate that the system can be used to do a proof of concept measurement but not as a drinking glass. In the future design section we discuss how we could possibly solve this.


We tested the diffusion of BSA, a similar-sized protein as BlcR, and glucose, a slightly larger molecule than GHB and SSA. We tested the diffusion through the filter after 1 and 6 hours and with and without shaking. The results can be seen in Figure 19 .

Figure 19. (a) Results from the protein assay in the membrane test. (b) Results from the glucose assay in the membrane test.

We observed limited diffusion of both protein and glucose. The protein was retained partly as needed for the system to function, but there was still diffusion which can affect the functioning and the safety of the system. The glucose did diffuse through the filter as intended but the diffusion was lower than expected. However, that would indicate that GHB can still pass from the beverage compartment to the compartment with the electrodes. Literature suggests lower diffusion of protein and higher diffusion of glucose [13] . In the future design section we discuss how we could improve the functioning of the filter.

Implementation

Proof of concept

The goal of our project is to detect GHB concentrations in beverages. As we could not use GHB due to legal reasons, we used the GHB analog SSA in most of our experiments. We verified using UTI measurements that the capacitance of the electrodes also changes with the addition of GHB. With the electrical circuit, we measured the capacitance after incubation of the electrodes with 1 µM DNA (Blc binding sequence: electrode A, scrambled DNA; electrode B) and 4 µM of BlcR. We used the button feature to negate the offset created by the inherent difference in capacitance values between the electrodes. We used the measured capacitance value as a baseline to measure the difference in capacitance after the addition of 150 µM SSA (Figure 20).

Figure 20. (a) Capacitance measurements with Arduino circuits. Baseline electrode with immobilized 1 µM DNA; electrode A: immobilized Blc operator sequence, electrode B: immobilized scrambled DNA, and 4 µM BlcR. Capacitance difference is measured in the presence of 150 µM SSA (Notebook 4, 27/09/2022).

After the addition of SSA, we measured a capacitance difference of 5.4*10-2 pF with electrode A and 0.0 pF with electrode B (Figure 20). The Arduino circuit was developed to turn on a light when a difference of 1.6*10-2 pF is measured and thus turned on the light. These results show that our Arduino system can indeed be used to detect SSA and most likely GHB. Due to problems resulting from leakage, malfunctioning electrodes and bad cable connections we were not able to detect SSA using the full device. In the future design section, we discuss possible ways of improving the prototype to achieve this.

User testing

We asked a bartender and a possible user of SPYKE to review, disassemble, and give feedback on the prototype. Pictures from the user tests are shown in Figure 21. As expected, a glass is very easy to use and only two remarks were given:

  1. The prototype is too large. As noted by both the the possible user and the bartender, the glass is too large and therefore difficult to hold. This makes the glass inconvenient to use.
  2. The prototype should be transparent. The bartender noted the glass should be transparent like regular glasses. Otherwise it would be very hard to identify what beverage is inside.
We were unable to make design changes to fulfill the wishes of the users as the size of the glass was limited by the electronics and the material by the available filaments. In the future design section we discuss possible ways of improving the prototype based on the input from the users

Lunch lecture SVLife
Lunch lecture Hooke
Figure 21. Pictures taken during the user tests.

Future design

The electrical circuit was able to detect SSA using, but the prototype still requires improvements before it can be used to prevent GHB spiking. Most importantly, the prototype should be further developed to also detect GHB inside the cup. The sensors performance should also be improved. Furthermore, the experiments conducted and user tests serve as a basis for more possible improvements. Lastly, the use of the device could also be expanded to detect other rape drugs. The following design changes should be implemented into the prototype.

  1. Improved sensor performance: The performance of SPYKE could be improved in multiple ways. First of all the sensor should be calibrated over a range of GHB concentrations to determine the average capacitance change to set the optimal threshold value. The button is a liability as it adds an extra step to using SPYKE. When produced the inherent difference between the electrodes should be measured from which a threshold should be installed before the glasses are handed out to the users. Experiments should also be conducted to improve filter performance to minimize the diffusion of protein and maximize the diffusion of SSA.
  2. Downsize and mass produce the glass: As identified during the user tests the prototype is too large, especially the diameter. The limiting factor for reducing the diameter is the size of the electronics. As explained by Dong Hoon Shin, the Arduino Nano should be replaced by a dedicated microcontroller to downsize the glass. This microcontroller should have all the capabilities to measure the capacitance and turn a LED, but nothing more. Thereby, decreasing the size of the electronics and subsequently the glass. The glass should subsequently also be mass produced to minimize cost, production errors and production time. The electronics could be embedded into the glass to allow for easier attachment between parts as the wires used at the moment are a liability.
  3. Change from 3D printing to rPET: As identified during the safe-by-design approach, 3D-printed materials could create a food safety risk. To avoid this, the material of the glass of SPYKE should be replaced with recycled PET (rPET). The top of the glass should be made from transparent rPEt. The bottom of the glass should not, to ensure the SPYKE glasses are indistinguishable from normal glasses to protect people without SPYKE cups. This is further elucidated in safe-by-design. The use of rPET would also improve the water tightness of the system as 3D-printed parts are prone to leakage.
  4. Sense multiple drugs: During the safe-by-design approach, we identified the potential risk of shifting from GHB to other drugs when spiking. To make SPYKE more comprehensive already identified electrical tests for other rape drugs ketamin [14] and benzodiazepines [15] should be integrated.
  5. Manual A user manual should be written for both nightlife facilities and users of the device. A QR code on the glass should be incorporated on the glass linking to the manual. The user manual should contain information on cleaning and distribution of the glasses for nightlife facilities and information on what to do when a high concentration of GHB is detected in a cup.

Downloads

The design of the prototype can be recreated for around €60. We share the files for the 3D printed pages and a recreation manual. In addition, all the software necessary to operate the read-out device is provided. For the full building guide click here. For the software to operate the Arduino, visit our gitlab repository.

!!! IT IS OF UTMOST IMPORTANCE BEVERAGES ARE NOT DRUNK AFTER BEING PRESENT IN THIS PROTOTYPE TO ENSURE FOOD SAFETY !!!

References

  1. Elliott, S. P. (2003, April). Gamma hydroxybutyric acid (GHB) concentrations in humans and factors affecting endogenous production. Forensic Science International, 133(1–2), 9–16. https://doi.org/10.1016/s0379-0738(03)00043-4
  2. Hubs.com. (n.d.) 3D printing with PLA vs. ABS: What's the difference? Retrieved 1 September 2022. Available at: https://www.hubs.com/knowledge-base/pla-vs-abs-whats-difference/
  3. www.sciencedirect.com. (n.d.) Ultrafiltration. Retrieved 10 September 2022. Available at: https://www.sciencedirect.com/topics/earth-and-planetary-sciences/ultrafiltration
  4. www.thermofisher.com. (n.d.) Dialysis Methods for Protein Research. Retrieved 10 September 2022. Available at: https://www.thermofisher.com/nl/en/home/life-science/protein-biology/protein-biology-learning-center/protein-biology-resource-library/pierce-protein-methods/dialysis-methods-protein-research.html
  5. www.nes-ips.com. (n.d.) Where and Why Are O Rings Used? Retrieved 20 August 2022. Available at: https://www.nes-ips.com/where-and-why-are-o-rings-used/#:~:text=O%20rings%20are%20used%20to,then%20compressed%20between%20two%20surfaces
  6. www.store.arduino.cc. (n.d.) Arduino Nano. Retrieved 12 August 2022. Available at: https://store.arduino.cc/products/arduino-nano
  7. www.analog.com. (2022) LTspice. Retrieved 2 October 2022. Available at: https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html
  8. en.wikiversity.org. (n.d.) Moving Average. Retrieved 12 August 2022. Available at: https://en.wikiversity.org/wiki/Moving_Average#:~:text=In%20statistics%2C%20a%20moving%20average,of%20finite%20impulse%20response%20filter
  9. nl.rs-online.com. (n.d.) 9V battery. Retrieved 10 October 2022. Available at: https://nl.rs-online.com/web/p/9v-batteries/7872281
  10. www.arduino.cc. (n.d.) analogread. Retrieved 12 August 2022. Available at: https://www.arduino.cc/reference/en/language/functions/analog-io/analogread/
  11. www.passive-components.eu. (2022) Leakage Current Characteristics of Capacitors. Retrieved 25 July 2022. Available at: https://passive-components.eu/leakage-current-characteristics-of-capacitors/ #:~:text=When%20a%20capacitor%20is%20charged%2C%20its%20leakage%20current%20drops%20with,capacitors%20have%20self%2Dhealing%20properties
  12. Heidari, A., & Meijer, G. C. M. (2010). A Low-Cost Universal Integrated Interface for Capacitive Sensors [PHD dissertation]. TU Delft.
  13. www.thermofisher.com. (2013) Separation Characteristics of Dialysis Membranes. Retrieved 10 September 2022. Available at: https://www.thermofisher.com/nl/en/home/life-science/protein-biology/protein-biology-learning-center/protein-biology-resource-library/protein-biology-application-notes/separation-characteristics-dialysis-membranes.html
  14. Fu, K., Zhang, R., He, J., Bai, H., & Zhang, G. (2019, October). Sensitive detection of ketamine with an electrochemical sensor based on UV-induced polymerized molecularly imprinted membranes at graphene and MOFs modified electrode. Biosensors and Bioelectronics, 143, 111636. https://doi.org/10.1016/j.bios.2019.111636
  15. Ashrafi, H., Hassanpour, S., Saadati, A., Hasanzadeh, M., Ansarin, K., Ozkan, S. A., Shadjou, N., & Jouyban, A. (2019, March). Sensitive detection and determination of benzodiazepines using silver nanoparticles-N-GQDs ink modified electrode: A new platform for modern pharmaceutical analysis. Microchemical Journal, 145, 1050–1057. https://doi.org/10.1016/j.microc.2018.12.017