Tri Circle Split Ring Resonator Shaped Metamaterial With Mathematical Modeling for Oil Concentration Sensing

In this research paper, the sensing capabilities of a tri circle split ring resonator shaped metamaterial is shown within the X-band frequency range for detection of various oil samples both theoretically and experimentally. Mathematical modelling was used to analyze the sensor efficiency for the various oil samples. A new sample holder has been made for the proposed sensor structure, and it showed significant performance. The simulated and measured transmission coefficient has been revealed in this study and monitoring the shift of resonance frequency and sensitivity of the metamaterial sensor for different oil samples. Since the dielectric constants of the oil samples differ, hence the resonance frequency is shifted. The obtained result revealed that the proposed sensor can detect a wide range of liquids, including (i) coconut and extra virgin coconut oils, (ii) olive and extra virgin olive oils, (iii) clean and waste engine oils, (iv) sunflower and canola oils. The resonance frequency has shifted to about 210, 230, 170, and 200 MHz for the above-mentioned oil samples, respectively. The sensitivity of the proposed MTM sensor is −83 dB (mg/L) which is a new analysis and quite significant as a potential sensor structure. Surface current and electric field distributions were used to deduce the sensing process. Since the recommended sensor is cheap and highly sensitive; so, it can be used in a variety of fields, such as detection of various liquids, microfluidic sensing, and industrial applications.


I. INTRODUCTION
Metamaterials (MTMs) are artificial materials made up of sub-wavelength resonant components capable of manipulating an electromagnetic (EM) field. It shows negative permeability and negative permittivity [1], but natural material does not show these properties. The MTMs physical features are widely reliant on the formed unit cells' design, shape, The associate editor coordinating the review of this manuscript and approving it for publication was Qi Luo . dimensions, and orientation. In recent years, research into the sensor applications of metamaterial has exploded. MTMs have an advantage over natural materials in that they can be engineered and tuned by manipulating their structural geometry and arrangement. Various applications of MTMs are biosensors [2]- [4], absorbers [5], [6], antennas [7], [8], energy harvesting, microwave sensors [9], [10], SAR reduction, and microwave lenses [11], [12]. MTMs provide several topologies; the most used topology is the split-ring resonator (SRR). SRR-based MTMs are used in various applications VOLUME 9, 2021 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ due to the adaptability and practical shape of the resonator. This research intends to make a tri circle split ring resonatorshaped metamaterial sensor for detecting various oil samples with high sensitivity. In the field of microwave sensors, the design of the MTM based sensor has gained enormous attention to the researcher [13]. Emerging technology allows us to create a new material with exceptional EM properties. The resonant frequency is an important feature of the MTM sensor's principle for material detection. An SRR based microstrip chemical sensor was demonstrated in [14] for the differentiating of ethanol and methanol solvents at 1.90 GHz operating frequency. The size of the sensor is small, and the quality factor is low. An oval-shaped sensor is presented in [15] for the glucose concentration measurement in an aqueous solution. The sensor's sensitivity was expected at 0.037 GHz per 30 mg/dL glucose solution. A triple ring resonator-based MTM sensor is described in [16] for fuel adulteration detection. A sample holder was attached with the designed structure in the frequency range of 8 to12 GHz in this analysis. The resolution and quality factor of this MTM sensor are high, but the resonant frequency shift is low due to the small change of the sample's dielectric constant. Some of the MTM sensors are described in [17], [18] based on the shift of the resonance frequency, but these values are low. An SRR-based MTM sensor is demonstrated in [19], for the microfluidic channel that is used to define resonance frequency shifts. Another MTM sensor demonstrated in [20] for the liquid's dielectric properties measurements, 60 MHz resonance frequency shifted in this analysis. An MTM sensor is demonstrated in [21] for the detection of different oils. The resonance frequency shifted 70 MHz for clean and dirty transformer oils and 50 MHz for olive oil and corn oil.
In [15], a left-handed MTM based sensor was presented for glucose sensing in the S-band. The mechanism of this sensor is the change of resonance frequency with the change of permittivity. Several designs of MTM sensors are described in [22]- [24] for the identity of various types of liquids. Resonance frequency has been shifted for the effect of the dielectric constant of the mentioned liquids. However, various approaches are developing with a great tendency for the wide applications of MTM sensors [25], [26]. It is noticeable that the MTM based microfluidic sensor is used for fuel adulteration sensing and content of ethanol in fuels [4], [27]- [29]. Kerosene is the most used ingredient for adulteration in diesel and petrol. In [30], oil pollution was described, where kerosene burning produces more noxious waste emissions, resulting from diesel and petrol engine damage. A meander line shape MTM-based sensor is demonstrated in [31] for the application of polypropylene sensing. It has desired sensitivity, compact size, small accuracy, and low-quality factor. G-shaped resonator-based MTM sensor is presented in [32] to detect liquid chemicals in the 8 to 12 GHz frequency range. The sensor works well to identify several liquids using frequency shift.
In this study, a new tri circle SRR shaped MTM with mathematical modelling is designed, analyzed, and fabricated for the detection of different oils within the X-band. To select the recommended resonator and size of the substrate, we analyzed the resonator size, split gap, radius, width, and the various dimension of the substrate. It is also noticeable that the suggested MTM sensor is cheap, high sensitivity, and suitable to be operated within X-band. Through the shifting of resonance frequency, the quality of the various samples has been detected. The recommended sensor is viable to be used in different applications, including industrial and liquid chemicals detection. The surface current and E-field distribution of the proposed sensor have been analyzed. The equivalent circuit has also been analyzed for the validation of the MTM structure.

II. DESIGN AND SIMULATION OF THE MTM SENSOR
The recommended MTM sensor has been presented in Figure 1. The electromagnetic high-frequency solver computer simulation technology (CST-2019) microwave studio has been used for the design and analysis of the sensor. The suggested MTM sensor consists of three circle SRR. The whole size of the MTM sensor is 22.86 × 10.16 mm 2 ; this dimension is selected for the adjustment of the X-band waveguide (WR90) since the guided opening of the WR90 waveguide is 22.86 × 10.16 mm 2 . Three suitable layers are used in this MTM sensor design; these are substrate, resonator, and sample holder (sensor layer). The thickness of the sample holder is 7 mm. Flame retardant-4 (FR-4) was used as a substrate, because of its minimal loss, low cost, and superior mechanical intensity. The relative permittivity and loss tangent of the FR-4 substrate is 4.3, 0.025, and its thickness is 1.5 mm. The resonator has been made of copper metal which is printed on both sides of the substrate. The thickness and conductivity of the copper are 0.035 mm, and 5 × 10 −8 S/m, in that order. The sample holder has been made by a different  layer of acrylic sheet. The dimension of the sample holder is 22.86 × 10.16 mm 2 which is the same dimension of the guided opening of the extended wave, WR90 waveguide, and MTM sensor. The sample holder is attached to the backside of the substrate, the thickness of the acrylic sheet for the sidewall is 1.5 mm, and the front and backside layer is 1 mm; the loss tangent and permittivity of the sensor layer is 0.004 and 2.4, respectively. The sample holder was used to keep the samples under test. There are three key layers to the MTM sensor. Figure 1 signifies the unit cell dimension; it is wellsuited for the waveguide (X-band), which is consistent with the investigational assessment. The sensor was created to work in the X-band frequency range because of the ease of access and fabrication of a sensitive arrangement that allows for a consistent sample holder. Table 1 shows the different parameter values of the used resonator.
The boundary conditions were applied in the simulation to see the transmission response (S 21 ). Intrinsically, using the WR-90 waveguide together with a corresponding sample holder to calculate S 21 . As shown in Figure 2, the boundary condition of the perfect electric conductor (PEC) was assigned to the x-and y-axes, while the z-axis (added space) along the propagation path is assumed to be free space. Due to the metallic structure of the side-wall waveguide, boundary conditions such as open space, periodic distribution, PEC/PMC, and PEC are suitable [32]. The basic reason for using PEC boundary conditions in the simulation study since we have realized experimental setup conditions. The simulation is carried out in the 8-12 GHz frequency range.
The novel idea of the proposed work is sensitivity analysis, mathematical modelling of the sensor, and designing a new sample holder for performance enhancement of the sensor. Also showed the better performance of the proposed sensor than the other reported MTM sensor ( Table 4 ). The sensitivity is an essential feature of the microwave sensor to analyze the dielectric characteristic. The sensitivity of the proposed MTM sensor is −83 dB (mg/L) which is a new analysis and quite significant as a potential sensor structure. We have also used mathematical modelling to justify the sensor's performance. Furthermore, a small change in the dielectric constant value causes a substantial shift in the resonance frequency, which proves the very good performance of the sensor. It's also worth noting that the proposed MTM sensor is inexpensive and suited for use in the X-band frequency. Since the recommended sensor is cheap and highly sensitive; so, it can be used in a variety of fields, such as detection of various liquids, microfluidic sensing, and industrial applications. Figure 3(a, b) depicts the various resonator designs and transmission responses (S 21 ) to them. In design 1, only 1 st ring has been used on the substrate, then the achieved S 21 magnitude is −27.79 dB at 11.38 GHz. To increase the electrical length, the 2 nd ring is added with the 1 st ring on the substrate, the obtained resonance4 frequency 8.23 GHz with a magnitude of −15.64 dB as shown in design 2. In design 3, the 3 rd ring is added with the 1 st ring on the substrate, the obtained S 21 value is −25.18 dB at 9.47 GHz. In designs 4, 2 nd , and 3 rd with the 1 st ring on the substrate, the obtained S 21 value is −25.18 dB at 9.47 GHz. In designs 4, 2 nd and 3 rd rings have been used on the substrate, the achieved magnitude of S 21 is −16.78 dB at 11.43 GHz. In the final design, 1 st , 2 nd , and 3 rd rings have been used on the substrate for increasing the sensing effect, the S 21 values are −15.37 dB at 8.34 GHz and −19.35 dB at 11.59 GHz. Since the two resonance frequencies and more sensing effects have been found from the tri circle SRR, hence the tri circle SRR has been selected for the proposed MTM sensor.
The tri (multiple) circle SRR shape has been proposed; since it can remove bianisotropic reactions and cross-polarization results in the dielectric mode as compared with a single SRR, and two SRR, because of its broadside character. Another advantage of using a tri circle SRR form is that it increases capacitive loading in the structure, which results in stronger resonance behavior. Furthermore, the proposed tri circle shaped SRR can achieve considerable miniaturization factors [48], [49]. Hence the tri circle SRR has been used for the suggested MTM sensor.
An SRR is one of the most extensively used non-natural magnetic materials [33], and it is made up of metal rings that are produced in a dielectric mode; and these shapes are circular, rectangular, spiral, omega, and others. Each form has its coupling impact based on its shape and EM qualities, such as side coupling in a circular structure, broadside coupler in a rectangular shape, and so on. Furthermore, the SRR exhibits bianisotropic effects and cross-polarization effects as the E-field stimulates the magnetic dipole moment. The bianisotropic performance and cross-polarization results are abolished if the metallic rings have a broadside character [34]- [36]. The tri circle SRR shape has been proposed; since it can remove bianisotropic reactions and crosspolarization impacts in the dielectric mode as compared with a single SRR, because of its broadside character. Another advantage of using a tri circle SRR form is that it increases capacitive loading in the structure, which results in stronger resonance behavior. Furthermore, the proposed tri circle shaped SRR can achieve considerable miniaturization factors [37], [38]. Hence the tri circle split ring resonator has been used for the proposed MTM sensor.
The revised circuit has given by including the characteristics impedance on the terminal part of ADS simulation as like CST simulation. The corresponding value is 377 at both terminals. In addition, an LC tank was added between the outer, and 2 nd ring due to high field concentration and surface current distribution. Therefore, the modified equivalent circuit is illustrated as in Figure 4(a). The inclusion of these new components does not change the overall S-parameter response significantly except a −10 dB magnitude decrease in S 21 . Overall, the resonance points remain the same, and bandwidth is approximately the same as the previous response. The LC tank circuit contributes to maintaining the overall response frequency. Therefore, a logical approach of equivalent circuit modification was performed to further clarify the proposed resonator. For each circle ring, a lumped component equivalent LC circuit has been analyzed to construct the geometry of the unit cell. Numerical analysis of these LC resonating components performed by commercially available Advanced Design System (ADS) software to approximate the RF nature. The estimation of the LC network was performed using the approximation mentioned in [15]. The equivalent circuit enhances the geometry estimation by presenting split gap and patch using LC network and each of the LC tank circuits mutually connected by a gap capacitor. After the simulation in ADS and CST, both S 21 performances were compared, as shown in Figure 4(b-d). The RF lumped component ADS response indicates that each potential resonance point exists with an approximate value of −38.11 dB at 11.32 GHz, −18.53 dB at 8.67 GHz, −25.20 dB at 9.01 GHz, and −38.75 dB at 11.63 GHz. All these points have a close approximation compared to each ring's corresponding CST response in the proposed unit cell. Therefore, a tri circle ring combination of the unit cell geometry would be a potentially  balanced and compact SRR choice to implement the sensing application.
The electric field distribution has been investigated to understand the designed sensor's mechanism. The electric field distribution provides information about the device's energy contained and losses. The electric field is created across the gap between the CSRR capacitive plate and the circled resonator during the resonances, making the region adjacent and inside the CSRR sensitive to dielectric changes. As a result, this region inside the CSRR can be used to test the dielectric characteristics of materials. In the empty sensor layer, the distributions are achieved at the resonance frequencies of 8.34 and 11.59 GHz.
The electric field strength is more concentrated in the resonator components, especially in the resonator's capacitive elements, as shown in Figure 5(a, b). At 8.34 GHz resonance frequency, the E-field distribution is effectively seen on the resonator, but at 11.59 GHz, it affects resonator parts with its surroundings due to the capacitive effect. Therefore, any small change in the sample's electrical characteristics in the sensor layer can be a sense in the proposed structure. Figure 6(a, b) shows the surface current distribution simulation graphs at the resonance frequency of 8.34 and 11.59 GHz. It is seen from the Figure; surface current is more accumulated on the inner and middle inductive circles due to the strong magnetic fields. In addition, that, more currents are disseminated in the resonator's left and right sides, regulating the electric and magnetic response, respectively. Parallel and anti-currents occurred at resonators, as seen in the Figure. Parallel currents were caused by the electric field, whereas anti-parallel currents were caused by the magnetic response to the applied EM field. The importance of resonators in resonant states is further explained in the Figure. The resonators are affected differently by each part of the resonators. The simulated surface current distribution for the recommended structure is a clear indicator of the presence of the electric dipole that generates the resonance phenomena. Figure 7 shows the reflection response (S 11 ) and transmission response (S 21 ) when the sample holder was empty. The Figure shows that the coefficients of reflection and transmission in the frequency spectrum of the X-band. This influence shows that when various oils are put in the sensor layer, then this is controlled within the 8-12 GHz frequency spectrum, the recommended MTM-based sensor form can be efficiently used to differentiate distinct oils. The S 21 values are −15.37 dB at 8.34 GHz and −19.35 dB at 11.59 GHz. As a result, various oil samples with electrically sensitive qualities have been selected to be examined in the preferred frequency range, both theoretically and experimentally.

III. PROPOSED SENSOR ANALYSIS WITH MATHEMATICAL MODELLING BASED ON SIMULATED AND MEASURED RESULTS
In this segment, using the FIT-based high-frequency EM solver CST microwave studio, a mathematical analysis is VOLUME 9, 2021   thick reservoir to be filled with different oils, where the transmitted EM wave's magnetic field is perpendicular to it in the z-axis. Port 1 and port 2 of the waveguide were attached to the front and back sides of the intended structure for numerical analysis and experimental testing of the transmission response (S 21 ). For the X-band waveguide, the guided opening dimensions are 22.86 × 10.16 mm 2 , which is the same size as the proposed structure.
To reveal the metamaterial property, the real and imaginary portions of the relative permeability (µ r ), relative permittivity (ε r ), refractive index (n), and impedance (Z ) of the sensor were acquired from the CST simulation, which is revealed in Figure 8(a-d).
The frequency range of negative effective parameters is listed in Table 2.
From Table 2, it is seen that the real part of the relative permittivity and relative permeability are negative in the frequency range of 10.98-11.57 GHz, which identifies the proposed resonator structure has potential MTM property.
The resonance frequency (f 0 ) of the suggested sensor has been affected by the resonator width, which is shown in Figure 9(a). When the sample holder was empty, the simulation has been done to show the shift of resonance frequency. In the proposed sensor, the resonator width has been changed, and observed the effect of the resonance frequency. There are five different widths of the resonator that has been used in this MTM based sensor, such as w = 0.4, 0.5, 0.6, 0.7, and 0.8 mm. It is seen that the f 0 is 8.20 and 11.21 GHz for the 0.4 mm resonator width and 8.46 and 11.79 GHz for the 0.8 mm resonator width. Since the resonator width is inversely proportional to the inductance, and inductance is inversely proportional to the f 0 , hence f 0 is increasing with increasing the width of the resonator. So, the sensor is properly obeying the relation of f 0 and inductance. The resonator width of 0.6 mm has been selected for the suggested MTM sensor. Figure 9(b) indicates that the impact of the different split gaps on the resonance frequency (f 0 ). When the split gap varies from 0.3 mm to 0.7 mm with an increment of 0.1 mm, then the f 0 has been seen 8.15 and 11.50 GHz for the 0.3 mm split gaps and 8.27 and 11.64 GHz for the 0.7 mm split gaps. Since the split gap is inversely proportional to the capacitance and capacitance is inversely proportional to the f 0 , so the value f 0 increases with increasing the split gap, i.e., the sensor properly obeying the relationship between f 0 and capacitance. The 0.5 mm split gap was used to fix the suggested MTM sensor design Figure 9(b) indicates that the impact of the different split gaps on the resonance frequency (f 0 ). When the split gap varies from 0.3 mm to 0.7 mm with an increment of 0.1 mm, then the f 0 has been seen 8.15 and 11.50 GHz for the 0.3 mm split gaps and 8.27 and 11.64 GHz for the 0.7 mm split gaps. Since the split gap is inversely proportional to the capacitance and capacitance is inversely proportional to the f 0 , so the value f 0 increases with increasing the split gap, i.e., the sensor properly obeying the relationship between f 0 and capacitance. The 0.5 mm split gap was used to fix the suggested MTM sensor design.
The effect of varied sample holder sizes on the f 0 is shown in Figure 9(c). If the size of the sample holder varies from 5 to 9 mm with an increment of 1 mm, then the lower f 0 values vary from 8.19 to 8.28 GHz and the higher f 0 value varies from 11.58 to 11.63 GHz. Here the size of the sample holder is inversely proportional to the capacitance, and capacitance is inversely proportional to the f 0 so, the value of f 0 is increasing with increasing the size of the sample holder i. e., the proposed sensor properly follows the f 0 and capacitance relation. In the proposed MTM sensor, the size of the sample holder 7 mm is selected.
The effect of unit cell orientation on the f 0 is shown in Figure 9(d). When the unit cell placed in the X-direction for simulation, got two values of f 0 , these are 8.34 and 11.59 GHz. If the unit cell is placed in the Y-direction, it also got two values of f 0 , these are 8.34 and 11.59 GHz which are the same as the X-direction. So, it can be said that there is no effect of unit cell orientation in the simulation. Figure 10 reveals the dielectric constant and loss tangent measuring setup for various oil samples. An open-ended coaxial probe was used to investigate the electrical properties of each sample. In this analysis, dielectric constant, and loss tangent measurements were carried out utilizing the N1500A dielectric probe kit. The probe kit was connecting with a power network analyzer (PNA)-L series vector network analyzer (VNA) N5224A in the 50 MHz to 43.5 GHz frequency range. The dielectric probe was calibrated using air and pure water at room temperature (25 • C) with well-known EM features in the frequency range of 8-12 GHz. In this frequency range, the dielectric constant and loss tangent of the different oils were obtained. The same process was repeated for each oil sample.     Figure 11(a-b). Figure 11(c) shows the oil insertion process in the sample holder. To fill oils, there is a gap on the upper side of the holder, which is closed with the same material to prevent adverse effects on the environment. By using the syringe, oils have been filled in the sample holder. The sample holder is filled with the 1.625 ml oil samples. To prevent contamination between the samples, each oil sample was placed in a different sample holder. The MTM sensor outputs are directly influenced by the dielectric constant of oils positioned inside the sample holder. Two waveguide ports and one extended guided wave attached to the coaxial cable linked by TNC female connector and N5227A PNA microwave network analyzer using the PNA-L series vector network analyzer (VNA) in the 10 MHz to 67 GHz frequency range is depicted in Figure 11(d). The MTM sensor is attached to the waveguide, which is shown in Figure 11(f). A calibration kit Agilent N4694-60001, was used to calibrate the VNA. Firstly, the fabricated metamaterial sensor is attached with the guided opening of the X-band waveguide. The capacitive parts relate to split gaps of the resonator are assembled as sensor layers.  The sample holder is placed on the suggested sensor between the waveguide and the extended guided wave, as shown in Figures 11(e). So, there is no direct contact between the oil sample and the metallic layer of the suggested sensor. This configuration was used to measure the transmission response (S 21 ) in the 8-12 GHz frequency range. For each sample, the same experimental procedure was followed.

A. MATHEMATICAL MODELLING OF THE METAMATERIAL SENSOR
During the sensing measurement, the waveguide port (WR90) considers the standard model for classical EM wave propagation. Medium variation during the propagation deals with relatively narrow bands of radiation; hence studying fixed frequency problems is quite reasonable to establish the mathematical modelling. During the measurement, the extended waveguide connected to port 1 passes through the air inside the guided path. Immediate medium changes such as substrate and sample oil interact with the propagated wave. After that, waveguide port 2 receives that wave passed through the resonator. Therefore, the transmission coefficient (S 21 ) of sensing element receives EM energy deviation using different samples during wave propagation. Assuming all timedependent components in both electric and magnetic fields as e −jωt using the following form [39] ∇ × E − jωµH = 0 (1) Inside the waveguide, permittivity, and permeability (ε air , µ air ) consider as a general parameter and the physical structure of propagation medium. The rectangular waveguide dielectric constant assumes a uniform distribution of dielectric constant ε (x) = (x 1 , x 2 , x 3 ) in each point of consideration. Periodicity in propagation direction inside the waveguide along the extended line should be the same for uniform distribution. Since the resonator structure is attached with the opening edge of the line, a variation of dielectric medium occurs and passes through two different mediums. The substrate permittivity and permeability symbolized as (ε s , µ s ) whereas for sample oil as (ε oil , µ oil ), respectively (Figure 12). Generally, EM wave propagation through the material with dielectric properties losses its magnitude and phase velocity. So, the absorbing capacity of the dielectric materials (substrate and oil samples in our case) characterizes the energy absorption and further synthesis describes the sensing characteristics of the proposed resonator mathematically. Besides, the structure has a finite extent in each point of consideration. Therefore, assuming non-conductive permittivity component for point x 1 rather than point x 2 , x 3 . Hence, EM wave before entering the resonator and sample considered as periodic form whereas during the dielectric medium propagation considered as quasi-periodic following well known 'Floquet-Bloch' theorem. So, for substrate and resonator field equations E αs = e −jαs E(s) and H αs = e −jαs H (s), for oil sample, E αo = e −jαs E(o) and H αo = e −jαs H (o). Now substituting these values with equations (1) and (2), we get [40] ∇ αs × E αs − jωµ s H αs = 0 (For sensing element ) ∇ αo × H αo + jωε o E αo = 0 (For sensing element ) (3) where modified '∇ operator' represents the differential component with complex quantity for respective medium variation. Now, the two different dielectric mediums lose the orthogonal property of EM waves due to the loss tangent and conical diffraction of the incident wave. The method of separation of variables for solving these partial differential equations is quite subtle and difficult for general description. Hence, the simplest solution approach is to find a second-order partial differential equation in 'n' variables by reducing equation (3) into second-order ordinary differential equations. So, Maxwell's equations reduce to simple scalar 'Helmholtz equation' [41] as αo + ω 2 ε αo µ αo u αo = 0 αs + ω 2 ε αs µ αs u αs = 0 where αo and αs represent second-order differential equation based on medium and u αo and u αs field component in Z-direction. Now, discretization of the equation (4) by applying the 'finite-difference time domain' method [42], [43] gives, where M represents the magnitude of the S-parameter with respect to resonator and oil sample. Extracting the S-parameter signal from another edge of the waveguide collect to analyze the sensitivity for the realization of the sensing element.
The mathematical modelling shows that any oil sample should include the transmission coefficient component jointly propagated through the measurement setup. So, in any arbitrary sample, the dielectric property and concentration is constant, but varying the sample would hamper these two parameters S 21 response. Figure 13 shows the overall response of all samples that have been analyzed during measurement. From this graph, it is seen that the f 0 gradually shifting over eight (8) samples signify the magnitude variation of transmission response (S 21 ) (dB). When the oil concentration is high, the magnitude becomes low, and the lower concentration shows a higher magnitude response. In this study, we have used eight (8)   are illustrated as −25.95 dB and −25.01 dB, respectively. Hence, we propose the definition of sensitivity of the resonator structure as S = G arg C oil (6) where G arg is an average gain variation for different oil samples and C oil is concentration variation of oil sample which is 0.28 mg/L in our case. The average gain variation was −23.375 dB. Therefore, the sensitivity approximately comes as −83.48 dB/(mg/L), which is quite significant as a potential prototype.

B. ANALYSIS OF THE COCONUT OIL AND EXTRA VIRGIN (EV) COCONUT OIL
Different types of oil may be of quality depending on their elements and natural conditions. Therefore, it was expected that the oil's electrical properties could be utilized as a feasible an instrument for investigating quality assurance and conception of their health benefit results. Coconut oil is extracted from dried and old coconuts and refined at high heat. Since refined coconut oil may have additional chemicals, so it is not recommended for external use, i.e., skin and hair. Extra virgin coconut oil is unrefined oil obtained by cold-pressed. Unrefined oil is that the least processed oil, and it contains Olive oil is a mixture that includes both cold-pressed and processed oils, while extra-virgin olive oil is made from VOLUME 9, 2021 pure, cold-pressed olives. Extra virgin olive oil is the least processed or refined type of oil. Extra-virgin olive oil is rich in monounsaturated fats and contains a small amount of vitamins E and K; it is also high in antioxidants, some of which have important health benefits. The dielectric constant (DK) and loss factor (LF) of these two oils are measured using an open-ended coaxial dielectric probe kit at the 8-12 GHz frequency range, shown in Figure 16. It is noticeable that the DK of olive oil is 2.54 and extra virgin olive oil is 2.63, where the LF of these two samples are 0.20 and 0.17 at the 8 GHz frequency.
The simulated and measured S 21 results for olive oil and extra virgin olive oil operate within X-band shown in Figure 17. The simulated magnitude of S 21 for olive oil is  reduce horsepower, reduce mileage, and shorten engine life. The dielectric constant (DK) and loss factor (LF) of the clean engine and waste engine oils are measured using an open-ended coaxial dielectric probe kit at 8-12 GHz frequency range, shown in Figure 18. It is noticeable that the DK of clean engine oil and waste engine oil is 2.33 and 2.24, where the LF of these two samples are 0.14 and 0.15 at the 8 GHz frequency. Figure 19 shows the simulated and measured results of the S 21 for clean engine and waste engine oil operating in X-band. The simulated magnitude of S 21

E. ANALYSIS OF THE SUNFLOWER OIL AND CANOLA OIL
The primary distinction between sunflower and canola oils is the type of fat they contain. Sunflower oil is high in monounsaturated and polyunsaturated fats, which help lower cholesterol, and canola oil is high in omega-3 fatty acids, a form of polyunsaturated fat that can help lower triglycerides. Firstly, the dielectric constant (DK) and loss factor (LF) of these two oils were measured using an open-ended coaxial dielectric probe kit at 8-12 GHz frequency range, shown in Figure 20. It is noticeable that the DK of sunflower oil and canola oil is 2.92 and 2.83, where the LF of these two samples are 0.11 and 0.16 at the 8 GHz frequency. Some ripples exist in the measurement transmission coefficient (S 21 ), which is the drawbacks of the presented work. But these ripples occurred due to the waveguide port's mutual coupling effect, and the fabrication tolerance of the prototype. Also, the mutual resonance effect of the transmitting and receiving ends of two waveguide ports always affect the readings and cause some variation in measurement data. We will try to optimize the drawbacks in future work. Furthermore, in the simulation result, all conditions are ideal, but in the measurement result, all are not ideal conditions; hence there are minor differences between these two results. So, it can be said that the proposed MTM sensor performs satisfactorily both theoretically and experimentally. Figure 22(a, b) shows the simulated and measured S 21 for the eight oil samples. It is noticeable that the transmission response (S 21 ) changes with changing the concentration and dielectric constant of oil samples. From Figure 22, we can easily distinguish between the clean engine oil and extra virgin olive oil or any two other oils. By using the suggested sensor, we can also know the transmission resonance frequency shifting and sensitivity of the oil samples. The close dielectric values are a big challenge in identifying the oil samples, but the proposed sensor identifies the oil samples by changing the resonance frequency.
oil samples. But when we have used the tri circle SRR in the MTM sensor, the sensor is working well, and one can easily detect the oil samples by the resonance frequency shifting and sensitivity of the samples. Since frequency is inversely proportional to the dielectric constant, hence there are some differences between extracted and reported dielectric constants due to different frequencies. Table 3 shows the comparison between extracted and reported dielectric constant values. Table 4 shows the comparison between the proposed study and other similar studies regarding the sensing materials, dielectric constant, resonance frequency shift, and sensor type. From references [17], [18], [21], [22] and [23], we can see that the resonance frequency shift of these references is slight, Besides, the resonance frequency shift is moderate, in the references [16], [24], [26], but the resonance frequency shift is high, in our proposed work. So, the performance of the proposed MTM sensor is better than other similar sensors, as stated in Table 4.

IV. CONCLUSION
A tri circle SRR shaped metamaterial sensor was effectively developed and fabricated such that it can be utilized for the detection of various liquids in the X-band frequency. The proposed sensor was shown to be capable of discriminating between various oil samples with ease. The mathematical modelling of the recommended MTM structure was also explained for sensitivity analysis. The dielectric constant depends on the oil concentration and frequency. The dielectric constant of the coconut oil and extra virgin coconut oil are 2.16 and 2.24, although the dielectric values are very close between these two oils; even after that, the resonance frequency has shifted to 210 MHz. The resonance frequency has shifted 230 MHz, though the dielectric constant values are nearby each other, 2.54 for olive oil and 2.63 for extra virgin olive oil. The dielectric constant values are 2.33 and 2.23 for clean engine oil and waste engine oil, respectively; cause of this small difference, the resonance frequency has shifted 170 MHz. To summarize, the proposed sensor is well suited for detecting sunflower and canola oil. The measured dielectric constant values for sunflower and canola oils are 2.92 and 2.83, respectively; for example, the resonance frequency shift between these two oils was around 200 MHz. The recommended structure is highly sensitive; the sensitivity is -83 dB (mg/L) which can be used in a variety of fields, such as detection of various liquids, microfluidic sensors, and industrial applications. The experimental and simulation findings were found to be in great agreement. It's also worth noting that the proposed MTM sensor is inexpensive and suited for use in the X-band frequency. The sensor's structure was successfully intended for real-time, long-lasting, and specific recognition of various samples. As a result, the proposed sensor could be very useful in a variety of industrial, liquid, and chemical detection applications.