While talking to potential users of our biosensor, it became clear to us that we need a portable hardware to enable the use of our biosensor. As mentioned in the theory section, we decided to measure the concentrations of the hormones with the Surface acoustic wave technique (SAW). So, we started to design a hardware prototype that enable future users an easy read-out of their hormone levels. For our prototype we needed to design several building blocks, one of the most important parts is the piezoelectric crystal. For our initial blueprint we chose a quartz crystal with a gold-coated sensing area and interdigital transducers (IDT). We chose the SAW chip from Turbé et al. for measuring HI virus particle concentration in blood [1] as a reference in our design process. Indeed, the application was similar to our needs. We contacted the co-authors D. Athey and H. Yatsuda to get detailed information about the composition of the SAW chip used in their experiments. After finishing our initial blueprint, we had several problems acquiring the necessary parts due to the pandemic. We realized that we had to adapt and had no choice but to work on our hardware theoretically. This meant we had to use modelling to simulate results from our hardware prototype to improve it according to the design-build-test cycle.

Figure 1 SAW sensor components. Quartz piezoelectric crystal, IDTs (interdigital transducer) are built on the crystal, the sensing film is located between the IDTs, the SAW Chip is inserted in a chip case which establish the connection to the electronics, the electronics are connected to the app to evaluate the data.

Quartz Substrate

The first important part of our hardware is the piezoelectric crystal. A possible candidate, and the most prominent one, is the quartz crystal. Quartz crystals are cheaper than other materials like lithium tantalate [1] and are being used for years, so they are well established for biosensor applications [2]. We wanted to orientate us on the chip in the publication “Towards an ultra-rapid smartphone- connected test for infectious diseases” so we contacted H. Yatsuda to get the dimensions of the quartz crystal. We choose the dimensions 3 mm x 5 mm x 0.5 mm (quartz wafer 4-inch and 0.5 mm thickness). We planned to order the quartz crystal from Roditi® but unfortunately, they would not have been able to deliver before our wiki freeze, so we had to make the hard decision to work out our hardware theoretically.

Figure 2 Dimensions of the piezoelectric quartz SAW crystal with IDTs and gold sensing film.

IDT fabrication

The IDTs (=interdigital transducers) are the part of the SAW chip that induce and receive the wave in the piezoelectric crystal, therefore transducing between an oscillating electrical signal and a physical surface acoustic wave. Each IDT is made up of two non-connected sets of interlocking "fingers" of gold coating on the surface of the piezoelectric crystal. The dimensions, the material and the excitation frequency are factors influencing the induced wave. They all depend on each other by which each component needs a specific value. The values our project is based on can be found in table 1.

Table 1 Measurements IDTs: finger pairs, material, frequency [1].

measurements	aperture A: 1 mm lambda λ: 20 um
finger pairs	80
material	gold
frequency	251.5 MHz

The aperture is the height of the IDT fingers, from one end to the end of the adjacent finger. The wavelength λ is the distance (center-to-center) between two adjacent, connected fingers. The width of one finger as well as the free space between interlocking fingers is equal to a quarter of λ. A schematic overview of the design and dimensions of the IDTs is given in figure 2.

Figure 3 Dimensions of IDTs, with commonly used symbols and selected values

The fabrication should have been done via a so-called lift-off procedure due to the equipment at hand. The protocol is based on the “conventional lithographic and lift-off technique” described by Illa et al. [3]. The setup requires a mask for the electrodes based on the values in table 1. The crystal needs to be cleaned and dried thoroughly before applying a UV photoresist is spun on followed by an image reversal process. Finally, the crystal is developed. The exact protocol can be found on the protocols page.

Sensing film

The sensing film is located between the IDTs on the crystal's surface. It is made out of gold, has a dimension of 1x1 mm and is 90 nm thick [1]. On this gold surface, the antibodies are immobilized. The sensing film is formed by evaporation. The crystal needs to be perfectly clean for adhesion of the gold layer, this is achieved by immersion into an acetone ultrasonic bath for 5 min, rinsing with water and drying with nitrogen. The area where no gold film should be applied is masked prior to this process. The disadvantage of gold is that it attracts dust strongly, so the chip must be protected from dust and should be used immediately. Also, the crystal needs to be perfectly clean otherwise the gold does not stick to it.

Connection SAW chip to electronics

There are different possibilities to establish the connection between the SAW chip and the circuit board. One possibility is the connection using so-called "Pogo pins". “Pogo pins” consist of a head component, a body, and an encapsulated spring. The connection is established by linking the head and the body [4]. The connection is not that stable, but still sufficient to be used for a momentary or a permanent connection. Another possibility is the connection with gold pads on the circuit board and gold pads linked to the IDTs on the crystal. The connection is established by pressing the contact pads on the crystal on the contact pads on the circuit board. The pads are pressed together by applying pressure with screws [5]. The third possibility to connect the IDTs to the circuit board would be wire bonding. The electrical connection is established by wires of different materials (gold, copper, silver, aluminium) which are processed with different temperature and pressure to establish a permanent connection [6]. We have chosen contact pads to establish the connection because of the possibility of a momentary connection so we could change the SAW chips. Also, a connection with contact pads is a very stable connection method and was used in a publication were they pointed out that the contact pad connection is a good option for connecting the IDTs to the circuit board [5]. Bonding establish a permanent connection, so it wasn’t suitable for our exchangeable chip approach.

Chip case

We designed a case to establish the connection between the SAW chip and the electronics. Furthermore, the application of the liquid (urine) on the sensing film can be implemented in this case. The contact pads located on the surface of the crystal establishes a connection between the IDTs and the circuit board. The circuit board in turn is connected to the electronics via SMA connectors. The liquid can be applied through a hole in the lid of the box. Silicone lips prevent the liquid from running to the IDTs. The urine is transferred onto the sensing film by a pipette. In further development a flow channel can be designed to applicate the urine onto the sensing film [2]. All components of the case can be printed with a 3D printer. Thumbscrews connect the different components of the case with each other.

Figure 4 Chip equipment, light grey: case, green: circuit board, gold: sensing film, screws: SMA connectors, dark grey: quartz crystal.

Figure 5 Chip equipment, plan view, light grey: case, green: circuit board, gold: sensing film, screws: SMA connectors, dark grey: quartz crystal.

Figure 6 Chip equipment, light grey: case, green: circuit board, gold: sensing film, screws: SMA connectors, dark grey: quartz crystal.

Figure 7 Chip equipment, side view, light grey: case, green: circuit board, gold: sensing film, screws: SMA connectors, dark grey: quartz crystal.

Analyzing the DDS

After explaining the general structure of our SAW chip, the next important aspect is thefrequency spectrum generated by our hardware. For SAW technology, it is essential to have constant frequencies that are as pure as possible to calculate very precisely the hormone concentration . SAW technology should only be used if the device is reliable and can generate frequencies without many interfering signals. We have thought about a device that can reliably generate frequencies in advance.

During our research we came across the direct digital synthesis (DDS) technology. This is described in the theoretical background to our hardware. To briefly explain DDS: A direct digital synthesizer (which is a microelectronic component with integrated circuits) is able to generate frequencies in any form. This could be e.g. a sine wave or a pulse wave. The possible settings for a DDS vary from model to model, but for nearly all devices these settings reach far beyond our project. First, the sine wave oscillations are of interest for us.

-PCB, needed to control the DDS, fit perfectly to the programmable pins of the Arduino Mega. Another advantage of the shield is the ability to adjust the frequencies directly on the shield with two buttons as well as to change certain settings of the synthesizer with an additional button. A little LCD shows the graphical user interface of the PCB.

To verify the functionality of the DDS and also the quality of the generated frequencies, we connected the DDS to an oscilloscope and visualized the frequencies. Afterwards we analyzed the frequency spectra of the DDS to determine if and how intense interference frequencies appear. The results are explained in the following section.

Frequency Spectra of the DDS

The frequency spectra of the AD9910 Arduino shield presented in the figures below are dependent on the clock frequency. We used a clock speed of 1.5 GHz which reduces the abundance and relative amplitude of interference in the frequency range of interest for our application. As one can see in Fig. 5 – Fig. 11 there is an amount of interference, small enough to be neglectable.

In comparison, at a clock speed of only 1 GHz, the frequency spectra of generated frequencies over 400 MHz comprise multiple disturbance peaks which could have a substantial impact on the excitation of the SAW chip’s piezoelectric crystal and thus on the reliability of the measurement. We therefore recommend using the shield’s option to set the clock frequency to 1.5 GHz. The difference in disturbance with the respect to the clock frequency is also visible in the following scope images.

Fig. 8 - Frequency spectra of a target frequency of 100 MHz (actual 103.5 MHz)

Fig. 9 - Frequency spectra of a target frequency of 200 MHz (actual 206.7 MHz)

Fig. 10 - Frequency spectra of a target frequency of 250 MHz (actual 258.3 MHz)

Fig. 11 - Frequency spectra of a target frequency of 400 MHz (actual 412.3 MHz)

Fig. 12 - Frequency spectra of a target frequency of 500 MHz (actual 514.6 MHz); one can see the disturbance peak in the low MHz range with approximately -50 dB.

Fig. 13 - Frequency spectra of a target frequency of 600 MHz (actual 617.0 MHz); one can see the disturbance peak at approx. 360 MHz with a level of approximately -47 dB.

To examine the functionality of the Arduino Shield we measured frequencies from 25 MHz up to 500 MHz. The frequencies can be set directly at the DDS. We observed a nearly linear factor which increases the target frequencies during this measurement (factor a1 = 1,028 at 1.5 GHz and factor a2 = 1.185 at 1 GHz). The factor is probably caused by the settings of the DDS, as it varies significantly or even disappears completely when selecting other clock frequencies. We decided to keep these settings anyway, because they create the most stable frequencies. This was important for the evaluation of the measuring data regarding the phase-shift. Therefore we used the derived factors to correct our measurement. The dependencies of the target and the actual frequencies generated by the AD9910 Arduino Shield are shown in figure 14.

Fig. 14 - Dependencies of the target and actual frequencies of the AD9910 Arduino Shield

At a target frequency of 450 MHz there is an irregularity in the trend of the measured frequencies with a clock speed of 1 GHz. In comparison the factor is considerably lower. Although we measured three different values to rule out the possibility of a measuring mistake, the factor is always lower than the other factors which is unusual. The irregularity is also shown in figure 14.

Gain adjustment

The amplifier can be adjusted by a software tool from the manufacturer which allows to select and change the gain of each channel separately. It is important to identify the amplitude generated by the DDS. In order to be able to give a reliable statement about the phase shift, it is necessary to work with an amplitude of approximately 0.5 V, since the components on our PCB operate with this voltage. Since the signal is doubled once for a reference measurement and once for our sensing chip, the amplitude decreases due to this operation and has to be amplified to approx. 0.5 V again.

To control the amplifier, it first needs to be powered by a 5 V supply voltage. We recommend exchanging the already soldered plugs for the supply voltage with pin headers, otherwise there is the risk of a short circuit. Moreover, the pin headers offer a direct connection to the GPIO of the Arduino, which delivers the supply voltage for the amplifier. Next the included USB cable can be connected to the computer.

After installing the software we first checked if the amplifier really increases the signal or whether there are difficulties with the configuration. In our case there was an incorrect configuration in the "Low Level View" menu. Two states required to be switched on 0, afterwards the amplifier could be controlled. The proper settings can also be found in the data sheet.

Fig. 15 - Low Level View of the LMH2832 Control Panel.

Fig. 16 - High Level View of the LMH2832 Control Panel.

In order to achieve an amplitude of at least 0.5 V, we have to adjust the gain to 13 which results in an amplification by the factor 10¹³. The settings are saved in the device's buffer memory, meaning that this step only requires a single adjustment. The images below illustrate the effect of the amplifier on the amplitudes.

The PCB

Fig. 17 - 3D-Model of the PCB, created with Altium Designer. The model is free to download below.

The PCB (Printed Circuit Board) is one of the most important parts of the technical realization of our project. Moreover, it is the only component that has been completely newly developed. In the following chapters we would like to explain briefly, why the board was developed in this way and what results were achieved.

As already mentioned in the human practice part, we worked together with the CITEC (Cluster of Excellence Cognitive Interaction Technology) of the Bielefeld University to define the technical implementation of our project idea and discussed the practical realization. The design of the board as well as the schematics were created by the working group "Cognitronics and Sensor Systems".

Various circuits would be suitable for measuring a phase shift. We discussed two possible realizations of the PCB.

One solution was to use an analog mixer. An exclusion criterion is the amplitude dependence of an analog mixer. This is problematic in the sense that at the time the circuit diagrams were developed, we did not have any knowledge of the amplitude of the frequency passing through the SAW chip. Therefore, this option was out of consideration. One can find more about the problems of an analog mixer in the Model section.

A second possibility, which was finally chosen, was the use of an XOR-gate. The functionality of the XOR is described in the Project/Hardware section. We attempted to capture the signal curve of the XOR at the contact points to visualize it on the scope. The signal curves of the XOR are shown in Fig. 18 and 19.

Fig. 18 - XOR curve on the scope.

Fig. 19 - XOR curve on the scope.

The Artificial Phase Shift

Since no real antibody-based phase shift could be created in the laboratory so far, it could not be evaluated with our SAW-technique. Therefore we had to create artificial phase shifts via the pin headers directly on the PCB. The areas marked in red in Fig. 17 can be connected in bridges, allowing a total of 16 different time shifts (in ps) to be simulated.

In order to cover the entire bandwidth of the DDS, we have generated the frequencies from 50 MHz to 600 MHz in 50 MHz steps for all 16 time shifts. We could detect the analog to digital converted voltage values on the serial console of the Arduino IDE. The results can be found in a table available for download below and in figures 18 and 19 underneath the text.

Download Table

Fig. 20 - PCB Pinheader Marked which could be connected with bridges to create an artificial time shift.

The results

Fig. 21 - Matlab Plot of the derived phase shift and time shift in 2D and 3D illustration.

The 3D plots illustrate the dependence of the time shift in [ps] or the phase shift in [°] and the frequency in [Hz] to the resulting voltages in [V]. At first it must be mentioned that the time shift was initiated with a specified track extension for a signal, which might cause the procedure to be incorrect. Furthermore, the phase shift is derived from the time shift, which can also be a source of mistakes.

We could prove by our measurement that our PCB delivers a voltage change at the output in case of a phase shift on the input, which we could detect by reading the analog output. However, it must be clarified that our PCB model is not able to detect a perfectly valid phase shift as well as voltage change.

To measure the precision of our board, further tests are necessary. The maximum mistake must be calculated to receive the percentage deviation. Moreover, the DDS generates the frequencies with a certain interference factor as well, which has already been analyzed in the section "Frequency spectra of the DDS".
Looking again at Figure 21, it becomes clear that e.g. there is supposed to be a minimum with a phase shift of 180°. But this point is located at a phase shift of 150°, which results a deviation of approximately 30°.

In conclusion, it can be said that even though there have been a number of systematic measurement deviations, we have developed a functional evaluation board that allows us to determine a phase shift from the analog output voltage value.

Files to download

To allow this project to be executed by further iGEM teams and possibly expanded, we have uploaded the files for the design of the PCB into our github repository. These files are available to download for free. Furthermore, one can find the source code for the Arduino Mega, which is used to read the PCBs analog output and additionally simulate phase shifts. We also uploaded the table of the results of the measurement as well as the matlab-plot so that the 3D plots can be reconstructed and viewed in three-dimensional.

Github-Link

The mobile device

3D Model of the Case

The housing for our measurement technology was designed with the program Fusion 360 from Autodesk. It consists of 2 parts: the actual box and the cover. The case dimensions are 77.5 mm (height) x 138.4 mm (width) x 207 mm (length).
One can recognize that there are honeycomb-like structures in the cover as well as in the actual box. These have not only an visual but also a functional reason. Because the DDS as well as the Arduino Mega, which operates the DDS, can get extremely warm (we measured 90°C surface temperature in some areas), the warm air needs to leave the case. The openings on the top and bottom of the case ensure sufficient air circulation.

The stl files of the 3D-modelled case can also be downloaded from the github.

Github-Link

Fig. 23 - 3D Model of the case for the various electronic components

Fig. 24 - 3D Model of the case for the various electronic components (Case)

Fig. 25 - 3D Model of the case for the various electronic components (Cover)

Sources

[1] V. Turbé et al., “Towards an ultra-rapid smartphone- connected test for infectious diseases,” Sci. Rep., 2017.
[2] K. Länge, B. E. Rapp, and M. Rapp, “Surface acoustic wave biosensors: A review,” Analytical and Bioanalytical Chemistry. 2008.
[3] X. Illa, O. Ordeig, D. Snakenborg, A. Romano-Rodríguez, R. G. Compton, and J. P. Kutter, “A cyclo olefin polymer microfluidic chip with integrated gold microelectrodes for aqueous and non-aqueous electrochemistry,” Lab Chip, 2010.
[4] Harwin, “Pogo Pins.” [Online]. Available: https://www.harwin.com/connectors-hardware/spring-contacts/pogo-pins/. [Accessed: 14-Oct-2020].
[5] S. Hohmann et al., “Surface Acoustic Wave (SAW) resonators for monitoring conditioning film formation,” Sensors (Switzerland), 2015.
[6] “Chip- und Wire-Bonding.” [Online]. Available: https://www.izm.fraunhofer.de/de/abteilungen/system_integrationinterconnectiontechnologies/arbeitsgruppen/qualifikations-_undprufzentrumqpz/chip-_und_wire-bonding.html. [Accessed: 14-Oct-2020].

Team:Bielefeld-CeBiTec/Hardware Results