Metamaterial sensor based on rectangular enclosed adjacent triple circle split ring resonator with good quality factor for microwave sensing application

In this article, a novel shaped metamaterial sensor is presented for the recognition of various oils, fluids, and chemicals using microwave frequency. The performance of the designed sensor structure has been studied both theoretically and experimentally, and it works well. A new sample holder for convenient operation is created and located just behind the designed structure. The results of this study performed better than those of prior liquids sensing studies. Various designs were explored using the Genetic Algorithm (GA), and it is embedded in the Computer Simulation Technology (CST) microwave studio, to optimize the optimal dimensions of the resonator. The suggested metamaterial sensor has a good-quality factor and sensitivity in both frequency shifting and amplitude changing. The resonance frequency shifted to 100 MHz between olive and corn oils, 70 MHz between sunflower and palm oils, 80 MHz between clean and waste brake fluids, and 90 MHz between benzene and carbon-tetrachloride chemicals. The quality factor of the sensor is 135, sensitivity is 0.56, and the figure of merit is 76 which expresses its efficient performance. Furthermore, the proposed sensor can sensitively distinguish different liquids by using the frequency shifting property. The study was carried out in three stages: dielectric constant (DK) measurement with the N1500A dielectric measurement kit, simulation of the structure, and experimental test study with the vector network analyzer. Since the recommended sensor has high sensitivity, good quality factor, and excellent performance, hence it can be used in chemical, oil, and microfluidic industries for detecting various liquid samples.

www.nature.com/scientificreports/ Design and analysis of the structure Figure 1a,b indicates the unit cell dimension and perspective position; it is well-suited for the waveguide (X-band). The whole length and width of the MTM sensor are 22.86 and 10.16 mm respectively, which is compatible with the X-band waveguide. The substrate material is flame retardant (FR-4) and the resonator is copper. On an FR-4 substrate with a thickness of 1.5 mm, a dielectric constant of 4.4, and a loss tangent of 0.02 the design is started. On both sides of the substrate, there were copper layers with a thickness of 0.035 mm. The sample holder is made of the polypropylene layer having a thickness of 1.5 mm, DK is 2.4, and the loss tangent is 0.004. The parametric analysis and the Genetic Algorithm (GA) approach were used to determine the required dimensions of the triple circle split SRR and the overall proposed structure. The GA approach was a built-in function inside the CST software 49 , which can be used to obtain the optimum result. A stochastic exploration technique acting on the tenets of pure genetic systems is a Genetic Algorithm. It operates a search to deliver the best possible result to an optimization issue for the condition function. The goal of achieving waveguide measurements in the simulation procedure for the designed structure with the effective size is to apply different boundary conditions. Because of the side-wall waveguide's metallic composition, it is appropriate to take the boundary requirements into justification, containing free space, periodic distribution, perfect electric conductor (PEC)/perfect magnetic conductor (PMC), and PEC. In the z-axis, the normal incident electromagnetic wave is used, whereas in the x-and y-axes, the perfectly electrified boundary condition is applied as shown in Fig. 2.
The design procedure of the suggested metamaterial sensor is depicted in Fig. 3a. The reflection coefficient ( S 11 ) for the various resonator designs of the MTM sensor are presented in Fig. 3b. All the circles had the same radius, and their value is 5 mm. When one circle is placed in the left corner of the selected substrate, then the value of S 11 is − 10.75 dB at 9.97 GHz frequency, which is shown in step 1. In step 2, the circle is placed in the middle position on the substrate, then the gain of S 11 is − 22.04 dB at 9.95 GHz frequency. When two circles are placed in the left and middle positions of the substrate, then the magnitude of S 11 is − 27.28 dB at 9.57 GHz frequency, which is shown in step 3. At 9.92 GHz, the magnitude of S 11 is − 20.94 dB when two circles are put in the left and right corners of the substrate, shown in step 4. At 9.50 GHz, the magnitude of S 11 is − 23.13 dB when three circles are put in the left, middle and right locations of the substrate, shown in step5. In the final step three circles are placed in the left, middle, and right positions of the substrate, and the rectangular has covered them, then the magnitude of S 11 is − 23.87 dB at 9.46 GHz frequency. Figure 4 represents the frequency versus S 11 graph for the different resonator structures. There are five resonator structures used to design the proposed sensor. The magnitude of S 11 is − 23.87 dB at 9.46 GHz for structure 1,   www.nature.com/scientificreports/   www.nature.com/scientificreports/ − 13.21 dB at 10.33 GHz for structure 2, − 9.29 dB at 11.63 GHz for structure 3, − 9.73 dB at 8 GHz for structure 4, and − 17.60 dB at 10.06 GHz for structure 5. The performance of structure 1 is better than others, so structure 1 is selected as a resonator pattern. At the resonance frequency of 9.46 GHz, the electric field distributions for the air are largely concentrated at the resonator surface shown in Fig. 5. The electromagnetic field is known to be able to travel across conductive cables. The field should be in transverse electromagnetic (TEM) mode, which means that the electric field is turned off. The distributions of electric field and surface current are also investigated to understand the designed sensor's theory. Changes in the electric field and surface current distributions provide information about the device's energy contained and losses. Figure 6 shows the circulation of surface current at the resonance frequency of 9.46 GHz. Surface current is more concentrated on the circle in the directions of clockwise and anti-clockwise. Furthermore, additional currents are dispersed on the left and right sides of the resonator, which regulates the electrical and magnetic responses. At the resonance frequency, the current concentration is equally in all circles. For the recommended structure, the existence of an electric dipole that creates a resonance event is demonstrated by the simulated surface current distribution. Figure 7a,b represents the capacitive and inductive segments and equivalent circuits of the designed MTM inspired sensor. The resonator can be represented by a total inductance ( L t ) and capacitance ( C t ). The resonator gaps denoted by the C g , so the resonator can behave like a LC model. Liquid samples with varying electrical characteristics can reduce the capacitance of the sensor layer (C s ) on the backside of the structure. The sensor layer's capacitance (C s ) is shown in the following equations 48 (1) C s = 4A − g C pul  www.nature.com/scientificreports/ here, A is the average resonator dimension, g is the gap of the split, and C pul is the capacitance per unit length. C pul can be computed as 51 where, c 0 is the free space light speed and Z 0 is the line impedance. So, the structure's total capacitance can be expressed as, here C 0 is the free space capacitance effect, C g is the capacitors for the gap, and C s is the effect of the sample's capacitance put within the sensor layer. The value of C s can be varied for various samples, because of variations in the complex dielectric permittivity characteristics that can be stated by 23 where, ε ′ sample is the real part and ε ′′ sample is the imajinary part of permittivity. The resonance frequency of the suggested structure can be determined using the following equation, here, L t is the structure's overall inductance. It is clearly shown that the f r is inversely proportional to the L t and C t of the recommended resonator. So, the inductor and capacitor play an important role in defining the sensitivity of the sensor. It is mentioned that the basic operation of the sensor structure is correlated to the interaction of the resonator and the sensor layer.

Parametric analysis of the metamaterial sensor
The reflection coefficient ( S 11 ) and resonance frequency f r are changing with the change of parameters i.e., resonator width (w), split gap (g), circle radius (r), sample holder thickness (SHT), substrate size (SS), and substrate material. In Fig. 8a the f r was more affected by the resonator width of the suggested sensor. In this MTM-inspired sensor, five different resonator widths have been used. If w = 0.5 mm, the S 11 value is − 19.76 dB  The f r is changes for the change of sample holder thickness (SHT), which is presented in Fig. 9b. For the SHT = 8, the magnitude of S 11 is − 21.08 dB at f r = 9.45 GHz. If the SHT = 9. 10, 11, 12 mm, then the magnitude of S 11 is − 22.35, − 23.87, − 26.18, and − 26.31 dB at f r = 9. 46, 9.46, 9.47, and 9.48 GHz, respectively. The sample holder's size was set to 10 mm.
The substrate size's (SS) effect on the f r is shown in Fig. 10a. When taking the SS is 22.36 × 9.66 mm 2 , the value of S 11 is − 24.40 at f r = 9.5 2 GHz. For the 22.86 × 10.16 mm 2 SS, the value of S 11 is − 23.87 at f r = 9.46 GHz. If the SS is 23.36 × 10.66 mm 2 , then the value of S 11 is − 23.17 at f r = 9.45 GHz. Since the substrate size is directly proportional to the dielectric constant and f r is inversely proportional to the dielectric constant hence when SS is increased f r is decreased. Excitation is given through waveguide WR90 (90WCAS with SMA female output) attached on both sides to measure the reflection response ( S 11 ). 22.86 × 10.16 mm 2 is the guided opening, so the SS of our designed MTM sensor is 22.86 × 10.16 mm 2 has been chosen. Figure Fig. 11b. From this circuit, it is seen that the capacitive effect is reduced because the circle split capacitors and inductors are on the same side and lumped effect. Hence the resonance frequency is slightly increased. Aso, when the proposed circle split rings pattern, is inverted the mutual coupling capacitor effect is decreased and the resonance frequency is increased which is shown in Fig. 11a. Figure 12 represents the dielectric constant (DK) and loss tangent (LT) measurement procedure for the different samples. The permittivity values and LT of the oils, fluids, and chemicals were determined using a calibrated Agilent network analyzer and the N1500A dielectric probe kit.

Results and discussions
Figure 13a-f depicts the suggested MTM sensor's entire measurement procedure. After the design process, the proposed sensor was developed using LPKF Laser and Electronics AG, Computerized Numerical Control (CNC) model Promat E33, and printed circuit board (PCB) machines, as shown in Fig. 13a,b. The container is filled with different liquid under test (LUT) samples and attached with the suggested structure. Each sample has been filled with a different holder to avoid contaminations between the samples. The Agilent N4694-60001 calibration kit was used to calibrate the PNA series vector network analyzer (VNA type N5227A). The coaxial cables link the VNA to the two waveguides. One waveguide is a transmitter, while the other is a receiver. With the sample holder, adaptor, waveguide, and sensor structure present, the S 11 was measured. First, the sample was injected into the sample holder, and then, in the target frequency range of 8-12 GHz, the S 11 was determined. Figure 13c reveals the liquid incorporation process in the sample holder, Fig. 13d shows the S 11 measurement system using waveguide, Fig. 13e,f represents the MTM sensor and sample holder connected with the waveguide. Figure 14 shows the schematic view of the microwave sensing measurement setup. Some fluctuations of oscillating shape have been seen on the VNA during the measurement. These fluctuations occurred due to the waveguides coupling effect and the MTM sensor fabrication tolerance during the prototype production of the FR-4 substrate layer. The dielectric constant of the substrate layer and testing environment have also influenced these fluctuations.  www.nature.com/scientificreports/ The measured and simulated reflection coefficient graph is depicted in Fig. 15, where the sample holder was filled with air. The value of S 11 is − 23.87 dB at 9.46 GHz in the simulation, and − 22.17 dB at 9.48 GHz in the measurement.
Study on the detection of olive oil and corn oil. Olive oil is a cooking oil obtained from olives, which are the fruit of the olive tree. It is most commonly used for high blood pressure, high cholesterol, and heart disease. This oil is also used in cooking and salad dressings. Corn oil is produced from the germ of the corn plant (maize). It is a refined vegetable oil that's commonly used in cooking, particularly deep frying. It is less expensive than the majority of other vegetable oils. Firstly, we have collected the olive and corn oils from the nearest super shop, then measured the DK and LT in the microwave lab using the dielectric measurement kit. After measuring the DK and LT of these oils, we have made the graphs. The graphs have been plotted using the obtained measured data, which is shown in Fig. 16. From the Figure, we can see that the value of DK is decreasing with increasing frequency, whereas the value of LT is increasing with increasing frequency. These phenomena show that, the DK and f r are inversely proportional, where the LT and f r are directly proportional. The measured DK is 2.56 and LT is 0.197 for olive oil at the 8-12 GHz range of frequency, where the DK is 3.07 and LT is 0.208 for the corn oil at that range of frequency.    www.nature.com/scientificreports/ The S 11 data for olive and corn oils have been plotted in Fig. 17. The simulated and experimental resonance frequencies are 9.45 GHz and 9.46 GHz for the olive oil with − 42.25 dB and − 36.59 dB magnitude, respectively. For corn oil, the magnitude values are − 31.51 dB at 9.35 GHz & − 28.67 dB at 9.35 GHz for the simulated and measured, respectively. The f r has been shifted to 100 MHz with − 10.74 dB magnitude change (simulated) and 110 MHz shifted with − 7.92 dB magnitude change (measured) for these two oils. By changing the resonance frequency, we can detect the presence of various oils. These findings show that despite their dielectric behaviour being very similar, the proposed MTM sensor accurately detects a variety of oils. The results of the measured and simulated are very similar. However, some of the fabrication and calibration errors produced minor discrepancies between them.

Study on the detection of sunflower oil and palm oil. Sunflower oil is a non-volatile oil obtained
from the seeds of sunflowers. It's widely utilized in the culinary industry as a frying oil and as a cosmetic emollient. It is used for constipation and lowering "bad" LDL cholesterol. Palm oil is an edible vegetable oil derived from palm tree fruit. Vitamin A deficiency, cancer, brain disorders, and aging are all prevented by this supplement. Firstly, we have collected the sunflower oil and palm oil from the nearest super shop, then measured the  www.nature.com/scientificreports/ dielectric constant (DK) and loss tangent (LT) in the microwave lab using the dielectric measurement kit. After measuring the DK and LT of these two oils, we have made the graphs. The graphs have been plotted using the obtained measured data, which is shown in Fig. 20. The DK and LT values are 2.93 and 0.099 for sunflower oil, whereas 2.81 and 0.149 for palm oil at 8 GHz frequency. The simulated and experimental S 11 data for palm oil & sunflower oil have been plotted in Fig. 21. The simulated and experimental resonance frequencies are 9.36 GHz and 9.37 GHz for the sunflower oil with − 27.54 dB and − 24.30 dB magnitude, respectively. For palm oil, the magnitude values are − 33.53 dB at 9.44 GHz and − 28.95 dB at 9.45 GHz for the simulated and measured, respectively. It is noticeable that the f r has been shifted to 80 MHz with − 5.99 dB magnitude change (simulated) and 80 MHz shifted with − 4.65 dB magnitude change (measured) for these two oils. By changing the resonance frequency, we can detect the presence of various oils. The results of the measured and simulated are very similar. However, some of the fabrication and calibration errors produced minor discrepancies between them.
Study on the detection of clean and waste brake fluids. Brake fluid is crucial to making sure that brakes run smoothly while driving the vehicle. It needs to be new and full to allow it to brake properly and keep safe. Sometimes, the brake does not work properly with the waste brake fluids, so, need to know the condition of the brake fluids. Firstly, we have collected the clean and waste brake fluid from the nearest motor repairing workshop, then measured the dielectric constant (DK) and loss tangent (LT) in the microwave lab using the dielectric measurement kit. After measuring the DK and LT of these two fluids, we have made the graphs. The graphs have been plotted using the obtained measured data, which is shown in Fig. 22   www.nature.com/scientificreports/ measured and simulated outcomes are in a nice contract with each other. However, a few of the fabrication and calibration errors produced minor discrepancies between them.

Study on the detection of benzene and carbon-tetrachloride.
Benzene is the most fundamental organic aromatic hydrocarbon and the parent compound of many aromatic compounds. It's a colorless liquid with a feature primarily utilized to make polystyrene. It's used to make plastics, resins, synthetic fibers, rubber lubricants, colors, detergents, medications, and pesticides. Carbon-tetrachloride is a synthetic compound that does not exist in nature. It's a clear liquid with a sweet smell that's detectable at low concentrations. It's utilized as a rubber solvent, a cleaning agent in the dry-cleaning business, and a solvent in the chemical and medicinal industries. Firstly, we have collected the benzene ad carbon-tetrachloride from the nearest chemical shop, then measured the DK and LT in the microwave lab using the dielectric measurement kit. After measuring the DK and LT of these two chemicals, we have made the graphs. The graphs have been plotted using the obtained measured data, which is shown in Fig. 24. The DK and LT values are 2.55 and 0.083 for benzene, whereas 2.51 and 0.079 for carbon-tetrachloride at 8 GHz frequency. The simulated and measured S 11 data for benzene and carbon tetrachloride have been plotted in Fig. 25. The simulated and experimental resonance frequencies are 9.38 GHz and 9.39 GHz for the benzene with − 24.79 dB and − 23.77 dB magnitude, respectively. For carbon tetrachloride, the magnitude values are − 28.41 dB at 9.46 GHz, − 25.62 dB at 9.48 GHz. The f r has been changed to 80 MHz with − 3.62 dB magnitude change (simulated) and 90 MHz shifted with − 1.85 dB magnitude change (measured) for these two oils. By changing the resonance frequency, we can detect the presence of various oils. The experimental and simulated findings are in a great deal with each other. However, some of the fabrication and calibration errors produced minor discrepancies between them.
When the concentration & DK of the samples are changed, the reflection response (S 11 ) varies. Minor discrepancies exist between simulated and measured results of S 11 due to the mutual coupling effect of the waveguide port, fabrication tolerance, and calibration inaccuracy. In addition, the mutual resonance effect of two  www.nature.com/scientificreports/ waveguide ports' transmitting and receiving ends will always affect the readings and produce minor variations in both readings. The accuracy of the sensor is the maximum difference that will exist between the actual value and the indicated value at the output of the sensor. Firstly, the dielectric permittivity of the liquid samples is measured by using the dielectric probe kit, then the reflection coefficients of the sensor are measured for the liquid samples. Then the linear curve fitting equation is formed by using the obtained resonance frequency with corresponding measured dielectric permittivity. The linear curve fitting equation is here f r is the resonance frequency. Now the measured dielectric constant of the olive oil is 2.46 at 9.46 GHz and the dielectric permittivity is obtained from the sensor is 2.48 at 9.46 GHz (using Eq. 6), again the dielectric constant of the clean brake fluid is 2.77 at 9.38 GHz and the dielectric permittivity is obtained from the sensor is 2.79 at 9.38 GHz which is very close to the measured value. So, it can be said that the accuracy of the sensor is superior.
The extracted sensitivities S(%) is defined as follows 34,52 : where, f is the initial frequency (when sample holder is empty), f 0 is the frequency when the sample holder is filled one after the material change and ε r is the permittivity value of the materials. Figure 26 represents the sensitivity vs permittivity curve. The quality-factor is a most important aspect of a metamaterial sensor to investigate the dielectric property. A common concern is that most MTM sensors suffer from low quality-factor and high measurement errors, and this, in turn, limits their usage in many applications. The quality-factor for the proposed sensor is, Q = f r /�f 53,54 ,  www.nature.com/scientificreports/ here f r is the resonance frequency and f is the + 3 dB bandwidth and it is dimensionless since the unit of both parameters are same. The difference between the two values of the independent variable at which the dependent variable equals 70.7% with respect to the minimal reflection value is known as the + 3 dB bandwidth. Again f = f h − f l , f h is the higher cutoff frequency, and f l is the lower cutoff frequency at 70.7% of its minimal reflection value. The Q-factor has been extracted by using this formula and the obtained value is 135. The maximum quality-factor was discovered via the olive oil simulation and experimental analysis. The quality-factor 135 has been found in the proposed metamaterial sensor.
As a quantitative description of the sensing performance characteristic, we calculated the figure of merit (FOM), which is defined as FoM = S × Q , where S is the sensitivity and Q is the quality factor 55 .
The sensitivity of the proposed sensor is S = 0.56 and the quality factor Q = 135 , hence the FoM = 0.56 × 135 = 75.6 ≈ 76 The frequency selectivity mechanism based on the concentration and purity of the liquid samples is used in the proposed sensor. The property of the sensor enables it to be "tuned" to respond better to certain frequencies than to others. The degree of selectivity of such filtering is sometimes specified as Q, which is the center or best frequency of a filter divided by its bandwidth. The proposed sensor identifies the concentration and purity of the liquid samples since the resonance frequency is shifted according to dielectric permittivity variation and the dielectric permittivity depends on the concentration and purity of the recommended liquid samples.
A comparison between measured DK and existing literature is indicated in Table 1. Table 2 shows the comparison among the proposed MTM inspired sensor and other sensors mentioned in the literature regarding the sensing materials, dielectric constant, the shift of f r , change of magnitude, Q-factor,and FOM. From references 20,23,27,28,[36][37][38]44 , and 47 we can see that the overall performance of our proposed work is better than the others.

Conclusion
Metamaterial sensor based on rectangular enclosed adjacent triple circle SRR shaped is presented for the recognition of various oils, fluids, and chemicals within the X band. The structure's performance has been studied both theoretically and empirically, and it has proven to be effective. This study was intended to provide a different approach to liquids sensing research. The whole dimension of the designed structure is 22.86 × 10.16 mm 2 , which is compatible with the X-band waveguide. The measured dielectric constant for olive oil is 2.56, and corn oil is 2.07, and the frequency has shifted 100 MHz with a reflection magnitude change of − 10.74 dB due to the variation of DK values between these two oils. The DK value for sunflower oil is 2.93, palm oil is 2.81, and the f r has shifted 80 MHz with a magnitude change of − 5.99 dB. The f r has shifted 70 MHz with a magnitude change of − 3.97 dB for clean and waste brake fluids, where the DK value is 2.78 for clean and 2.69 for waste brake fluids. The f r has shifted 90 MHz with an amplitude change of − 3.62 dB between benzene and carbon tetrachloride, where the DK value is 2.55 for benzene and 2.51 for carbon tetrachloride. The suggested MTM sensor has a good quality factor and high sensitivity in both frequency shifting and amplitude changing. The quality factor and FOM of the sensor are 135, and 76 which expresses its efficient performance. Furthermore, the proposed sensor can sensitively distinguish different liquids by using the frequency shifting property. For various oils, fluids, and chemicals, the measured and simulation findings are very similar. If the dielectric constant of the sample is changed, the obtained findings indicate that this idea can readily be applied to diverse applications of electrochemical sensing. Since the recommended sensor has high sensitivity, good-quality factor, excellent performance, and low price, hence it can be employed in a variety of applications, including industrial and liquid chemical detection.