Verification of Quartz Crystal Microbalance Array Using Vector Network Analyzer and OpenQCM

Received Jan 13, 2018 Revised Mar 15, 2018 Accepted Mar 30, 2018 Quartz Crystal Microbalance (QCM) is a device that allows non-destructive measurements of r in situ reaction activities. In this article, an array comprising of six 3MHz QCM sensors in an array were characterized using a vector network analyzer and OpenQCM, a portable measuring instrument that measures change in resonance frequency. Measurements of S21 transmission characteristics using the vector network analyzer provides the resonance frequency and can also be used to derive the RLC equivalent electrical circuit values of a resonant two-port network based on the Butterworth-Van Dyke model. In this work, Rm, Lm, Cm and Co were obtained via curve-fitting of the measurement results to the simulated results. Measurements were done in triplicates to verify reproducibility for all 6 sensors. For comparison, measurements were also done using a portable, open-source instrument, OpenQCM. The OpenQCM instrument directly measures changes in resonance frequencies, making it ideal for biosensing experiments, which correlate changes in mass with changes in resonance frequencies. Comparison between resonance frequency measurements using VNA and OpenQCM exhibit low percentage difference 0.2%. This QCM sensor array has the potential of conducting real-time, point-of-care analyses for detection of biological molecules.


INTRODUCTION
Over the past two decades, intensive research on biosensors have been carried out in the development of label-free quantification and point-of-care medical diagnostics [1]- [4]. Biosensors can be defined as analytical devices that capture biological responses associated with a particular disease (biorecognition element) in the form of detectible electrical signals (transduction method). Some typical examples of bio-recognition elements are cells, enzymes, biomolecules, aptamers, and antibodies [5]. The transduction method can translate these bio-recognition elements into quantifiable signals. Several types of transducers have been employed over the years as biosensors; e.g. electrochemical [6], [7], mass-sensing (mechanical) [8], [9], optical [10], [11], and field-effect transistors [12].
There is an increasing demand for specific, low cost, portable and highly sensitive biosensors. Point-of-care (POC) biosensor systems provides a comprehensive solution as it provides label free detection, real-time monitoring, that is easily portable which is also relatively cheap [1], [13]- [22]. Technically, POC devices can be classified as portable systems that are separate from other miniaturized platforms due to their  Zainuddin) 85 ability to perform analyses in the field. These systems can also be implemented such that it integrates with the multiple existing lab processes using microfluidic systems that allows detection using various types of transduction methods and real-time result acquisitions [23]. Among the different types of transduction methods, optical methods such as surface-plasmon resonance or SPR, fluorescence, and optical cavity resonators were reported to be some of the most sensitive [10], [11], [21]. SPRs measure surface-based chemical reactions that correspond to refractive index changes, thus giving change to the optical signal. SPRs are however, impractical for POC applications due to its expensive equipment which also require specific technical expertise to operate [13]. Recent work have shown interest on analyses of other transduction methods such as electrochemical detection and mass-sensing. These methods have exhibited several advantages for instance, ease of operation, low cost, and minimal operating procedures [6]- [9]. Furthermore, electrochemical detection like electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) utilizes screen printed electrodes (SPEs) which are commonly used in the biosensing field [24]- [28]. This method detects impedance or current changes in electrode surface owing to transport or surface reactions [23]. However, this method is principally categorized as a destructive process since it applies direct current (DC) and alternate current (AC) through a system, hence deteriorating the electrode surface, specifically in SPE applications. Although the permanent change on the electrode surface is in nano-range scale, it still destabilizes the current or impedance of the system. Hence, non-uniform results are observable if the same SPE is used in another application. For this reason, most SPEs are disposable, where the electrodes are only meant for single-use applications for specific target detection and specific transduction methods like EIS and CV [29]. Although SPEs are low-cost POC devices, it will increase operational cost and time especially when it involves establishing multiple testing or results verification.
Apart from the electrochemical methods, mass-sensing or Quartz Crystal Microbalance sensors (QCMs) are also widely used as it is a non-destructive and ultrasensitive sensor. These sensors only apply very low amplitude alternate currents (AC) to the top and bottom electrode surfaces of piezoelectric substrates, typically AT-cut Quartz crystals. This results in a substantially low current being applied to measure the chemical interactions on the electrode's surface without physically modifying it, ensuring longer usage and reproducibility of results. For certain experiments, QCMs can also be cleaned using chemistry solutions or ultrasonic treatment, allowing it to be reused. Due to this, QCMs have been widely used for biochemistry detection [30]- [37]. QCM sensors can evaluate the changes in mass and density-viscosity of complex biological responses on the surface of its crystal sensor based on changes in the resonance frequency [38]. The resonance frequency of the propagating acoustic wave through the surface of the quartz will reduce due to mass change [39]- [41]. In most common cases, these sensors can even detect frequency shifts of 1 Hz by using high frequency oscillators (in the MHz range) [42] allowing it to detect masses in the sub-ng level. The QCMs generally have low spurious bulk signals and good temperature stability. The wave mode in a QCM is bulk thickness shear mode (TSM), which allows for operation in both dry and liquid environments. The mass-sensing method has been used (via Sauerbrey theory) in various fields such as environmental analyses, security monitoring, food safety and medical diagnostics [31], [43].
In recent years, an increasing amount of literature has been reported on the development of QCM arrays in biosensing applications [44]- [46]. These works aid users to execute parallel and simultaneous multi-target analyses. It will significantly simplify tedious procedures of manipulating concentrations, data range, and different target molecules by having multiple arrays on a single disk. Furthermore, it can be integrated with microfluidic systems for continuous real-time experiments without significant error in quality factor [44]. A QCM that operates in aqueous environment can detect density-viscosity and acoustic impedance effects, making it a highly sensitive device. Placement of several QCMs in an array allow multiple parallel measurements, improving the throughput of the device. Arrangement of these sensors in array need to take into consideration several aspects such as cross-talk between sensors, energy loss due to liquid damping and other factors [47]- [49]. One key point for the design of these sensors is that the Q-factor of each sensor has to be high to overcome liquid damping. An optimal gap between the sensors is also necessary to avoid frequency interference, hence decreasing the Q-factor error for each sensor.
This paper focuses on the simulation and verification of the QCM array using vector network analyzer (VNA) and an open source portable measuring instrument, (OpenQCM). In this work, we simulate and fabricate six 3MHz of QCM sensors on a single wafer. Design details of the QCM have been reported previously in [50]. Initially, a classical Butterworth-van-Dyke (BVD) model was used to calculate the impedance parameters (RLC) of each QCM sensor. The results were then used to simulate a resonance frequency analyses using ADS software. Comparison between the theoretical and experimental resonance frequency analyses was performed next. This report is organized into six sections with the first section introducing the biosensor and the significance of POC in diagnostic applications. Section 2 comprises the underlying theories behind the proposed QCM sensor as Section 3 deals with the ADS simulation model. The  Figure 1 shows the top and cross sectional views of the device together and operation as a mass sensor. This biosensor comprises two deposited metal on the top and bottom of a quartz crystal. The variables r and s indicate the radius of working electrode and centre-to-centre distance of two adjacent QCM sensors, respectively. The diameter of quartz substrate is 76mm, quartz thickness, h = 500µm and sensor radius, r = 3.30mm were used in this work. The centre-to-centre distance of QCM, s is was set to 6mm to minimize frequency interference. An AC signal is applied between the top and bottom AT-quartz electrodes to generate acoustic wave energy and produce resonance. When used for biosensing applications, frequency shifts will be induced when biological materials are placed on the top electrode due to additional mass. Placement of six sensors in an array allows multiple detection in a single platform.

Equivalent circuit models
The QCM can be electrically represented using an equivalent circuit based on the classical Butterworth-van-Dyke (BVD) circuit as shown in Figure 2 [51]- [52]. This theory is established from a onedimensional analyses of a piezoelectric resonator as described in [52]. The RLC parameters of the equivalent circuit can be divided into motional components, R m , L m and C m which are derived from the resonance operation of the QCM and an additional parallel static capacitor (C 0 ). This additional C 0 contributes to the dielectric energy storage because the oscillation crystal is established in between the two electrodes. The motional resistance, R m represents dissipation or energy loss during resonance. The motional impedance, L m and motional capacitance, C m correspond to the mass of the resonator and coupling coefficient, respectively. The equations for these motional elements are as expressed in Eq. (1)-(4). Table 1 summarizes the material properties used to perform this calculation.

EXTRACTION AND SIMULATION OF RLC EQUIVALENT CIRCUIT
In this section, a computer-aided synthesis program (ADS software) was used in order to determine the RLC equivalent circuit parameters of the QCM sensor. The calculated values of RLC parameters are predicted using Equation (1)-(4) as described in Section 2. The RLC equivalent circuit is next simulated using ADS in a two-port network to obtain its S21 or transmission parameters. Figure 3 shows the S2p simulation setup in which port 1 and port 2 correspond to input and output ports, respectively. An AC frequency domain analysis is done from 2.77MHz to 3.77MHz. A step size of 5kHz is set to obtain highaccuracy simulations. Tuning and optimization methods are done next to extract relevant physical parameters. The simulation results are compared and tuned to experimental measurement results to obtain the RLC parameters of the QCM sensor.

EXPERIMENTAL WORK
In this section, two different equipments were used to measure the resonance frequency of the QCM array. In the first experiment the QCM sensors were measured using a vector network analyzer (VNA), Agilent E5061A. In the second experiment, the OpenQCM device was used to measure the resonance frequency. The OpenQCM is essentially an Arduino microcontroller having ATMega32U4 processor with 16 MHz clock speed. The frequency of the vibrating quartz crystal was measured using the FreqCount algorithm developed by Paul Stoffregen from PJRC. The software on the OpenQCM is open source and can be customized for different applications. The main motivation to utilize the OpenQCM, is its portability and its ease of use compared to VNAs, which are unsuitable for field measurements and require additional data processing for mass measurements. Usage of OpenQCM does have some limitations however, as it can only detect QCMs with resonance Qs of larger than 1000. Figure 4 shows the connection setup of the QCM array to the VNA via a printed circuit board (PCB). The PCB was fabricated with the input and output ports that uses two SMA connectors. Short 3-cm wires were soldered at both input and output ports to enable a connection to the QCM sensor using clip connectors. In order to obtain high-accuracy measurements, AC frequency domain analyses were set from 2.77MHz to 3.77MHz with a step size of 5kHz.  Figure 5 indicates the experimental setup of the QCM sensor array using the OpenQCM device. The input and output ports were connected via direct clip connectors. The OpenQCM device was connected to the computer using a USB connector. The measurement records the QCM's resonance frequency, duration of the experiment and the current temperature. In this work, the duration was set to 300s in order allow sufficient time for each QCM to achieve stable resonances before the measurements are recorded.  Figure 6 shows the measured resonances for QCM1, QCM3 and QCM6. Measurements were done in triplicates for each QCM. Measurement results are reproducible. The average resonance frequency of 3.281MHz ± 0.004MHz was recorded in this work.  Figure 7 shows the both the simulated ADS and measured S21 transmission characteristics. From the graph, it can be seen that both results are in good agreement with a resonance frequency of close to 3.281MHz. The extracted equivalent circuit parameters are detailed in Table 2.  Figure 5. Extracted characteristics were simulated using the calculated values shown in Table 2 via Equation (1-)(4) and next fine-tuned to match the measurement results. Figure 8 shows the experimental results for all QCMs. From the results, a stable transmission frequency of 3.290±0.004MHz for 6 QCMs array in a constant temperature of 31 C. These results complied with the resonance frequency measurement using VNA with only a small tolerance of 0.3% and 0.4% respectively. Table 3 summarizes the analyses of resonance frequency measurement in terms of theoretical and experimental methods as discussed previously in this section.

CONCLUSION
An array of biosensors comprising of 6 QCMs on a single wafer have been fabricated and characterized in this work. Measurements were done using both a vector network analyzer (VNA) and a portable instrument, OpenQCM have been discussed. Measurements of S21 transmission characteristics using the VNA can also be used to determine the RLC equivalent electrical circuit values of R m , L m , C m and C o . Measurements were done in triplicates to verify reproducibility for all 6 sensors. Next, measurements using a portable, open-source instrument, OpenQCM were made. The OpenQCM instrument directly measures changes in resonance frequencies, making it ideal for biosensing experiments, which correlate changes in mass with changes in resonance frequencies. Comparison between resonance frequency measurements using VNA and OpenQCM exhibit low percentage difference 0.2%. This sensor has the potential of real-time analyses and parallel monitoring or detection of biological molecules. When integrated with the OpenQCM, this sensor has the potential of performing on the field analyses due to its portability and ease of use.