Meander-Line Slot-Loaded High-Sensitivity Microstrip Patch Sensor Antenna for Relative Permittivity Measurement

A high-sensitivity microstrip patch sensor antenna (MPSA) loaded with a meander-line slot (MLS) is proposed for the measurement of relative permittivity. The proposed MPSA was designed by etching the MLS along the radiating edge of the patch antenna, and it enhanced the relative permittivity sensitivity with an additional effect of miniaturization in the patch size by increasing the slot length. The sensitivity of the proposed MPSA was compared with that of a conventional rectangular patch antenna and a rectangular slit (RS)-loaded MPSA, by measuring the shift in the resonant frequency of the input reflection coefficient. Three MPSAs were designed and fabricated on a 0.76 mm-thick RF-35 substrate to resonate at 2.5 GHz under unloaded conditions. Sensitivity comparison was performed by using five different standard dielectric samples with dielectric constants ranging from 2.17 to 10.2. The experiment results showed that the sensitivity of the proposed MPSA is 6.84 times higher for a low relative permittivity of 2.17, and 4.57 times higher for a high relative permittivity of 10.2, when compared with the conventional MPSA. In addition, the extracted relative permittivity values of the five materials under tests showed good agreement with the reference data.


Introduction
Accurate characterization of the complex permittivity of a material is very important in various applications, such as wireless communication, biomedical, healthcare, agriculture, chemistry, and the food industry [1]. There exist various permittivity measurement methods, and they are classified into non-resonant and resonant methods. Recently, the resonant methods with planar resonators, such as split ring resonator (SRR) and complementary SRR (CSRR) structures, have been widely used because of their simple geometry, small size, ease of fabrication, and low cost [2][3][4]. Microstrip structures have been extensively applied for complex permittivity measurement because of their advantages compared to other structures such as low weight, low profile, low cost, conformance to platform surface, ease of fabrication, and compatibility with integrated circuits. SRR-based and CSRR-based sensors in the microstrip transmission line as a band stop filter have been extensively investigated [5][6][7][8][9][10][11][12][13][14][15][16][17][18]. An interdigital capacitor-based SRR structure coupled with a microstrip transmission line was used as a microstrip resonant sensor at 2.45 GHz to enhance the relative permittivity sensitivity of dielectric materials [19]. A CSRR structure based on an interdigital-capacitor-shaped defected ground structure was introduced at 1.5 GHz for high-sensitivity relative permittivity characterization [20]. The measured sensitivity of the proposed sensor was two times higher for a low relative permittivity material at 2.17, measured by the shift in the resonant frequency of the input reflection coefficient (S 11 ) when the planar dielectric material under test (MUT) was placed above the patch as a superstrate. The three MPSAs were designed and fabricated on a 0.76-mm-thick RF-35 substrate for the first resonant frequency at 2.5 GHz under unloaded conditions. Full-wave simulations were performed using CST Microwave Studio. Figure 1 shows the geometries of a conventional inset-fed rectangular MPSA, an RS-loaded MPSA, and the proposed MLS-loaded MPSA. The corresponding S 11 characteristics of the three MPSAs are shown in Figure 2.

MLS-Loaded MPSA Design
Sensors 2019, 19, x FOR PEER REVIEW 3 of 16 sensitivity was measured by the shift in the resonant frequency of the input reflection coefficient (S11) when the planar dielectric material under test (MUT) was placed above the patch as a superstrate. The three MPSAs were designed and fabricated on a 0.76-mm-thick RF-35 substrate for the first resonant frequency at 2.5 GHz under unloaded conditions. Full-wave simulations were performed using CST Microwave Studio. Figure 1 shows the geometries of a conventional inset-fed rectangular MPSA, an RS-loaded MPSA, and the proposed MLS-loaded MPSA. The corresponding S11 characteristics of the three MPSAs are shown in Figure 2. The conventional inset-fed rectangular MPSA was designed to resonate at fr1 = 2.5 GHz under unloaded conditions by using the equations in [37] with an RF-35 substrate (ε r = 3.5, tan δ = 0.0018, h = 0.76 mm), as shown in Figure 1a [34]. The calculated width and length of the rectangular patch were W1 = 40.0 mm and L1 = 31.9 mm, respectively. The width and length of the 50 ohm microstrip feed line were wf1 = 1.66 mm and lf1 = 24.5 mm, respectively, whereas those of the inset were wis1 = 2.8 mm and lis1 = 9 mm, respectively. The width and length of the ground plane are Wg = Lg = 80 mm. The MUT will be placed above the patch as a superstrate for loaded conditions, as shown in Figure 1a. Two other higher-order resonant frequencies exist at fr2 = 4.03 GHz and fr3 = 4.584 GHz.

MLS-Loaded MPSA Design
Next, an RS was etched along the radiating edge of the patch on the opposite side of the microstrip feed line, as shown in Figure 1b, and two resonant frequencies were generated at fr1 = 2.5 GHz and fr2 = 3.465 GHz with a frequency ratio of fr2/fr1 = 1.39 [34,38]. The frequency bandwidth of the first resonant frequency for a voltage standing wave ratio (VSWR) < 2 was 2.496-2.503 GHz (0.28%), which is quite reduced compared to the conventional MPSA (2.490-2.510 GHz, 0.8%) because of the The conventional inset-fed rectangular MPSA was designed to resonate at f r1 = 2.5 GHz under unloaded conditions by using the equations in [37] with an RF-35 substrate (ε r = 3.5, tan δ = 0.0018, h = 0.76 mm), as shown in Figure 1a [34]. The calculated width and length of the rectangular patch were W 1 = 40.0 mm and L 1 = 31.9 mm, respectively. The width and length of the 50 ohm microstrip feed line were w f1 = 1.66 mm and l f1 = 24.5 mm, respectively, whereas those of the inset were w is1 = 2.8 mm and l is1 = 9 mm, respectively. The width and length of the ground plane are W g = L g = 80 mm. The MUT will be placed above the patch as a superstrate for loaded conditions, as shown in Figure 1a. Two other higher-order resonant frequencies exist at f r2 = 4.03 GHz and f r3 = 4.584 GHz.
Next, an RS was etched along the radiating edge of the patch on the opposite side of the microstrip feed line, as shown in Figure 1b, and two resonant frequencies were generated at f r1 = 2.5 GHz and f r2 = 3.465 GHz with a frequency ratio of f r2 /f r1 = 1.39 [34,38]. The frequency bandwidth of the first resonant frequency for a voltage standing wave ratio (VSWR) < 2 was 2.496-2.503 GHz (0.28%), which is quite reduced compared to the conventional MPSA (2.490-2.510 GHz, 0.8%) because of the size reduction of the RS-loaded patch. The third resonant frequency, f r3 , existed slightly above 5 GHz. The width and length of the ground plane and the width of the feed line were the same as the conventional MPSA in Figure 1a. The width and length of the rectangular patch were scaled down to W 2 = 31.8 mm and L 2 = 25.4 mm, respectively, in order to make the first resonant frequency 2.5 GHz.
The size reduction was about 20.4%, compared to the conventional MPSA. The length of the feed line was slightly increased to l f2 = 27.3 mm. The width and length of the RS were w rs1 = 1 mm and l rs1 = 29.8 mm, respectively. The distance between the radiating edge and the thin rectangular slot was w o1 = 1 mm. The width and length of the inset were w is2 = 2.3 mm and l is2 = 12 mm, respectively. Note that the location of the RS was chosen because the electric field concentration on the patch was highest at this position as shown in Figure 3b. size reduction of the RS-loaded patch. The third resonant frequency, fr3, existed slightly above 5 GHz. The width and length of the ground plane and the width of the feed line were the same as the conventional MPSA in Figure 1a. The width and length of the rectangular patch were scaled down to W2 = 31.8 mm and L2 = 25.4 mm, respectively, in order to make the first resonant frequency 2.5 GHz.
The size reduction was about 20.4%, compared to the conventional MPSA. The length of the feed line was slightly increased to lf2 = 27.3 mm. The width and length of the RS were wrs1 = 1 mm and lrs1 = 29.8 mm, respectively. The distance between the radiating edge and the thin rectangular slot was wo1 = 1 mm. The width and length of the inset were wis2 = 2.3 mm and lis2 = 12 mm, respectively. Note that the location of the RS was chosen because the electric field concentration on the patch was highest at this position as shown in Figure 3b.  Thirdly, an MLS was loaded along the radiating edge of the patch, as shown in Figure 1c. The shape of the proposed MLS looked like a single-rectangular-ring CSRR with L-shaped slots appended at both sides of the split gap, and its total length was much longer than that of the RS. In this case, the frequency bandwidth of fr1 for a VSWR < 2 was further reduced to 2.498-2.502 GHz (0.16%) with an increased quality factor. The dimensions of the patch were scaled down to W3 = 22.0 mm and L3 = 17.6 mm, which were reduced by about 44.9% compared to the conventional MPSA. A quarterwavelength transformer was used on the feed line to match the 50 ohm input impedance. The dimensions of the MLS and other geometric parameters to make the first resonant frequency 2.5 GHz were as follows: wrs2 = 1 mm, lrs2 = 20.0 mm, wo2 = 1 mm, g1 = 1 mm, lrs3 = 9.5 mm, g2 = 1 mm, lrs4 = 9.5mm, wqt = 0.6 mm, lqt = 18.3 mm, wis3 = 1.5 mm, lis3 = 4 mm, wf3 = 1.66 mm, lf3 = 31.2 mm, and Wg = Lg = 80 mm. In this case, the second resonant frequency was fr2 = 3.233 GHz, located relatively close due to the increased slot length compared to the RS-loaded case. The third resonant frequency existed at fr3 = 4.584 GHz. The relationship between the first resonant frequency and the geometric parameters of the MLS could be derived based on the resonant frequency variation obtained from the simulation, and it was as follows:  size reduction of the RS-loaded patch. The third resonant frequency, fr3, existed slightly above 5 GHz. The width and length of the ground plane and the width of the feed line were the same as the conventional MPSA in Figure 1a. The width and length of the rectangular patch were scaled down to W2 = 31.8 mm and L2 = 25.4 mm, respectively, in order to make the first resonant frequency 2.5 GHz.
The size reduction was about 20.4%, compared to the conventional MPSA. The length of the feed line was slightly increased to lf2 = 27.3 mm. The width and length of the RS were wrs1 = 1 mm and lrs1 = 29.8 mm, respectively. The distance between the radiating edge and the thin rectangular slot was wo1 = 1 mm. The width and length of the inset were wis2 = 2.3 mm and lis2 = 12 mm, respectively. Note that the location of the RS was chosen because the electric field concentration on the patch was highest at this position as shown in Figure 3b.  Thirdly, an MLS was loaded along the radiating edge of the patch, as shown in Figure 1c. The shape of the proposed MLS looked like a single-rectangular-ring CSRR with L-shaped slots appended at both sides of the split gap, and its total length was much longer than that of the RS. In this case, the frequency bandwidth of fr1 for a VSWR < 2 was further reduced to 2.498-2.502 GHz (0.16%) with an increased quality factor. The dimensions of the patch were scaled down to W3 = 22.0 mm and L3 = 17.6 mm, which were reduced by about 44.9% compared to the conventional MPSA. A quarterwavelength transformer was used on the feed line to match the 50 ohm input impedance. The dimensions of the MLS and other geometric parameters to make the first resonant frequency 2.5 GHz were as follows: wrs2 = 1 mm, lrs2 = 20.0 mm, wo2 = 1 mm, g1 = 1 mm, lrs3 = 9.5 mm, g2 = 1 mm, lrs4 = 9.5mm, wqt = 0.6 mm, lqt = 18.3 mm, wis3 = 1.5 mm, lis3 = 4 mm, wf3 = 1.66 mm, lf3 = 31.2 mm, and Wg = Lg = 80 mm. In this case, the second resonant frequency was fr2 = 3.233 GHz, located relatively close due to the increased slot length compared to the RS-loaded case. The third resonant frequency existed at fr3 = 4.584 GHz. The relationship between the first resonant frequency and the geometric parameters of the MLS could be derived based on the resonant frequency variation obtained from the simulation, and it was as follows: Thirdly, an MLS was loaded along the radiating edge of the patch, as shown in Figure 1c. The shape of the proposed MLS looked like a single-rectangular-ring CSRR with L-shaped slots appended at both sides of the split gap, and its total length was much longer than that of the RS. In this case, the frequency bandwidth of f r1 for a VSWR < 2 was further reduced to 2.498-2.502 GHz (0.16%) with an increased quality factor. The dimensions of the patch were scaled down to W 3 = 22.0 mm and L 3 = 17.6 mm, which were reduced by about 44.9% compared to the conventional MPSA. A quarter-wavelength transformer was used on the feed line to match the 50 ohm input impedance. The dimensions of the MLS and other geometric parameters to make the first resonant frequency 2.5 GHz were as follows: w rs2 = 1 mm, l rs2 = 20.0 mm, w o2 = 1 mm, g 1 = 1 mm, l rs3 = 9.5 mm, g 2 = 1 mm, l rs4 = 9.5 mm, w qt = 0.6 mm, l qt = 18.3 mm, w is3 = 1.5 mm, l is3 = 4 mm, w f3 = 1.66 mm, l f3 = 31.2 mm, and W g = L g = 80 mm. In this case, the second resonant frequency was f r2 = 3.233 GHz, located relatively close due to the increased slot length compared to the RS-loaded case. The third resonant frequency existed at f r3 = 4.584 GHz. The relationship between the first resonant frequency and the geometric parameters of the MLS could be derived based on the resonant frequency variation obtained from the simulation, and it was as follows: where L 3 is the length of the rectangular patch, f r1 is the first resonant frequency, c is the speed of light in free space, ε reff is the effective relative permittivity, ∆L is the extension of the patch length due to the fringing effects, a 0 is the coefficient to compensate for the shift in the first resonant frequency due to the inset between the patch and the quarter-wavelength transformer, a 1 is the coefficient related to l rs2 , a 2 is the coefficient related to 2g 1 , a 3 is the coefficient related to 2l rs3 , a 4 is the coefficient related to 2g 2 , and a 5 is the coefficient related to 2l rs4 . ε reff and ∆L can be calculated by using the equations in [37], which are ε reff = 3.3 and ∆L = 0.3538 mm. The calculated values of the coefficients were as follows: a 0 = 0.4248 × 10 −3 , a 1 = 0.2355, a 2 = 0.5627, a 3 = 0.1511, a 4 = 0.3031, and a 5 = 0.2612. Figure 3 shows the electric field distributions of the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs at the first resonant frequency of 2.5 GHz. We can see that the electric fields are widespread along the two radiating edges of the patch for the conventional MPSA. When RS is loaded along one of the radiating edges, the electric fields are concentrated along the RS, as shown in Figure 3b. For the proposed MLS-loaded MPSA, the electric fields were more concentrated near the MLS with the reduced area. Therefore, the size of the MUT can be extensively reduced, compared to the conventional MPSA.

Sensitivity Comparison
In this section, the sensitivity of the proposed MLS-loaded MPSA is compared with the conventional and the RS-loaded MPSAs by measuring the shift in the first resonant frequency of the input reflection coefficient. Figure 4 shows the S 11 characteristics of the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs when the MUT was placed above the patch as a superstrate. The MUT's relative permittivity (ε r ) was varied from 1 to 10 in increments of 1 with a zero loss-tangent, as in the lossless case. The width and length of the MUT are the same as those of the ground plane, and its thickness was chosen to be 1.6 mm because the thickest of the substrates available from Taconic Inc., which were used as the MUTs, was around 1.6 mm. For the conventional MPSA, the first resonant frequency of S 11 moves from 2.5 GHz to 2.312 GHz when ε r increases from 1 to 10, whereas it shifts from 2.5 GHz to 1.854 GHz for the RS-loaded MPSA. The first resonant frequency of the proposed MLS-loaded MPSA moves from 2.5 GHz to 1.585 GHz when ε r increases from 1 to 10. Table 1 summarizes the first resonant frequencies of the S 11 responses with a different relative permittivity of the MUT for the three MPSAs.
where L3 is the length of the rectangular patch, fr1 is the first resonant frequency, c is the speed of light in free space, εreff is the effective relative permittivity, ΔL is the extension of the patch length due to the fringing effects, a0 is the coefficient to compensate for the shift in the first resonant frequency due to the inset between the patch and the quarter-wavelength transformer, a1 is the coefficient related to lrs2, a2 is the coefficient related to 2g1, a3 is the coefficient related to 2lrs3, a4 is the coefficient related to 2g2, and a5 is the coefficient related to 2lrs4. εreff and ΔL can be calculated by using the equations in [37], which are εreff = 3.3 and ΔL = 0.3538 mm. The calculated values of the coefficients were as follows: a0 = 0.4248ⅹ10 -3 , a1 = 0.2355, a2 = 0.5627, a3 = 0.1511, a4 = 0.3031, and a5 = 0.2612. Figure 3 shows the electric field distributions of the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs at the first resonant frequency of 2.5 GHz. We can see that the electric fields are widespread along the two radiating edges of the patch for the conventional MPSA. When RS is loaded along one of the radiating edges, the electric fields are concentrated along the RS, as shown in Figure 3b. For the proposed MLS-loaded MPSA, the electric fields were more concentrated near the MLS with the reduced area. Therefore, the size of the MUT can be extensively reduced, compared to the conventional MPSA.

Sensitivity Comparison
In this section, the sensitivity of the proposed MLS-loaded MPSA is compared with the conventional and the RS-loaded MPSAs by measuring the shift in the first resonant frequency of the input reflection coefficient. Figure 4 shows the S11 characteristics of the conventional, the RS-loaded, and the proposed MLSloaded MPSAs when the MUT was placed above the patch as a superstrate. The MUT's relative permittivity (εr) was varied from 1 to 10 in increments of 1 with a zero loss-tangent, as in the lossless case. The width and length of the MUT are the same as those of the ground plane, and its thickness was chosen to be 1.6 mm because the thickest of the substrates available from Taconic Inc., which were used as the MUTs, was around 1.6 mm. For the conventional MPSA, the first resonant frequency of S11 moves from 2.5 GHz to 2.312 GHz when εr increases from 1 to 10, whereas it shifts from 2.5 GHz to 1.854 GHz for the RS-loaded MPSA. The first resonant frequency of the proposed MLS-loaded MPSA moves from 2.5 GHz to 1.585 GHz when εr increases from 1 to 10. Table 1 summarizes the first resonant frequencies of the S11 responses with a different relative permittivity of the MUT for the three MPSAs.    In order to compare the sensitivity of the proposed MPSA with that of the conventional and the RS-loaded MPSAs, we used the following definitions [2,9,10,20,31,34]: where Δfr is the shift in the first resonant frequency of the three MPSAs, fru is the first resonant frequency of the three MPSAs for unloaded conditions, frl is the first resonant frequency of the three MPSAs for loaded conditions, PRFS is the percent relative frequency shift of the first resonant frequency of the three MPSAs, PRFSE is the enhancement in the percent relative frequency shift of the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, S is the sensitivity of the first resonant frequency of the three MPSAs with respect to the relative permittivity of the MUT, εru is the relative permittivity of the MUT for unloaded conditions, εrl is the relative permittivity of the MUT for loaded conditions, and SE is the sensitivity enhancement of the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs compared to the conventional MPSA.
Based on the results shown in Figure 4 and Table 1, the sensitivity performance of the three MPSAs was analyzed using Equations (2) to (6). First, Δfr of S11 for the three MPSAs is plotted in Figure 5a. Figure 5b shows the PRFS of the three MPSAs. The PRFSE for the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, is presented in Figure 5c. Next,   In order to compare the sensitivity of the proposed MPSA with that of the conventional and the RS-loaded MPSAs, we used the following definitions [2,9,10,20,31,34]:

Antenna Type
where ∆f r is the shift in the first resonant frequency of the three MPSAs, f ru is the first resonant frequency of the three MPSAs for unloaded conditions, f rl is the first resonant frequency of the three MPSAs for loaded conditions, PRFS is the percent relative frequency shift of the first resonant frequency of the three MPSAs, PRFSE is the enhancement in the percent relative frequency shift of the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, S is the sensitivity of the first resonant frequency of the three MPSAs with respect to the relative permittivity of the MUT, ε ru is the relative permittivity of the MUT for unloaded conditions, ε rl is the relative permittivity of the MUT for loaded conditions, and SE is the sensitivity enhancement of the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs compared to the conventional MPSA.
Based on the results shown in Figure 4 and Table 1, the sensitivity performance of the three MPSAs was analyzed using Equations (2)- (6). First, ∆f r of S 11 for the three MPSAs is plotted in Figure 5a. Figure 5b shows the PRFS of the three MPSAs. The PRFSE for the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, is presented in Figure 5c. Next, S of the first resonant frequencies for the three MPSAs is plotted in Figure 5d. Figure 5e presents the SE of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA.
We observe from Figure 5d that sensitivity S is not constant as a function of relative permittivity. For example, when the permittivity of the MUT is ε r = 2, frequency shift ∆f r in the first resonant frequency of the conventional MPSA is 0.032 GHz, whereas for the RS-loaded and the proposed MLS-loaded MPSAs it is 0.116 GHz and 0.187 GHz, respectively. PRFS in the first resonant frequency of the conventional MPSA is 1.28%, whereas for the RS-loaded and the proposed MLS-loaded MPSAs it is 4.64% and 7.48%, respectively. Therefore, PRFSE for the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, are 3.63 and 5.84. In this case, S is the same as ∆f r because the difference in the relative permittivity (∆ε r ) is 1. SE values of the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, are 3.63 and 5.84. It is worthwhile to note that PRFSE is the same as SE, and it can be used to estimate the sensitivity enhancement when the resonant frequencies of the MPSAs are different, because the sensitivity enhancement cannot be used for different resonant frequencies.
When the permittivity of the MUT is increased to ε r = 3, ∆f r in the first resonant frequency of the conventional MPSA is 0.  Next, in order to determine the relationship between PRFS of the first resonant frequency in the proposed MLS-loaded MPSA and the relative permittivity of the MUT, a curve-fitting tool in SigmaPlot was used and a fifth-order polynomial function was chosen for the fitting function.
The PRFS of the simulated first resonant frequency for the proposed MPSA when the MUT's relative permittivity (εr) was varied from 1 to 10 in increments of 1 was used to derive Equation (7). Figure 6 compares the simulated and curve-fitted relative permittivity of the MUT as a function of PRFS. Errors between the simulated and curve-fitted relative permittivity at the input data points were kept to less than 0.1%. Hence, we can conclude that the sensitivity of the proposed MLS-loaded MPSA at the first resonant frequency is 4.87 to 6.00 times higher than the conventional MPSA for relative permittivity values ranging from 2 to 10.
Next, in order to determine the relationship between PRFS of the first resonant frequency in the proposed MLS-loaded MPSA and the relative permittivity of the MUT, a curve-fitting tool in SigmaPlot was used and a fifth-order polynomial function was chosen for the fitting function.
The PRFS of the simulated first resonant frequency for the proposed MPSA when the MUT's relative permittivity (ε r ) was varied from 1 to 10 in increments of 1 was used to derive Equation (7). Figure 6 compares the simulated and curve-fitted relative permittivity of the MUT as a function of PRFS. Errors between the simulated and curve-fitted relative permittivity at the input data points were kept to less than 0.1%.  Next, in order to determine the relationship between PRFS of the first resonant frequency in the proposed MLS-loaded MPSA and the relative permittivity of the MUT, a curve-fitting tool in SigmaPlot was used and a fifth-order polynomial function was chosen for the fitting function.

Experiment Results and Discussion
Prototypes of the three MPSAs were fabricated on an RF-35 substrate (ε r = 3.5, tan δ = 0.0018, h = 0.76 mm), as shown in Figure 7. The S 11 characteristics of the fabricated antenna were measured using an Agilent N5230A network analyzer, and photographs of the experiment setup are shown in Figure 8. Five different standard dielectric MUTs from Taconic, Inc., were tested for the sensitivity comparison. The MUTs had relative permittivity ranging from 2.17 to 10.2, and their relative permittivity, loss tangent, and thickness from the data sheet in [39] are summarized in Table 2.
First, the S 11 characteristics of the three MPSAs were simulated with the five MUTs in Table 2, as shown in Figure 9. The length of the MUTs was slightly reduced to 75 mm because there is a protruding part in the SMA connector for soldering the input port of the fabricated antenna, as shown in Figure 8b. For the conventional MPSA, the first resonant frequency for S 11 moved from 2.465 GHz for the TLY-5A MUT (ε r = 2.17) to 2.312 GHz for the RF-10 MUT (ε r = 10.2). In the RS-loaded MPSA, the first resonant frequency shifted from 2.369 GHz for TLY-5A to 1.852 GHz for RF-10, whereas it moved from 2.285 GHz for TLY-5A to 1.577 GHz for RF-10 in the proposed MLS-loaded MPSA. Table 3 shows the first resonant frequencies of the simulated S 11 responses for the three MPSAs.

Experiment Results and Discussion
Prototypes of the three MPSAs were fabricated on an RF-35 substrate (εr = 3.5, tan δ = 0.0018, h = 0.76 mm), as shown in Figure 7. The S11 characteristics of the fabricated antenna were measured using an Agilent N5230A network analyzer, and photographs of the experiment setup are shown in Figure  8. Five different standard dielectric MUTs from Taconic, Inc., were tested for the sensitivity comparison. The MUTs had relative permittivity ranging from 2.17 to 10.2, and their relative permittivity, loss tangent, and thickness from the data sheet in [39] are summarized in Table 2.
First, the S11 characteristics of the three MPSAs were simulated with the five MUTs in Table 2, as shown in Figure 9. The length of the MUTs was slightly reduced to 75 mm because there is a protruding part in the SMA connector for soldering the input port of the fabricated antenna, as shown in Figure 8b. For the conventional MPSA, the first resonant frequency for S11 moved from 2.465 GHz for the TLY-5A MUT (ε r = 2.17) to 2.312 GHz for the RF-10 MUT (ε r = 10.2). In the RS-loaded MPSA, the first resonant frequency shifted from 2.369 GHz for TLY-5A to 1.852 GHz for RF-10, whereas it moved from 2.285 GHz for TLY-5A to 1.577 GHz for RF-10 in the proposed MLS-loaded MPSA. Table  3 shows the first resonant frequencies of the simulated S11 responses for the three MPSAs.

Experiment Results and Discussion
Prototypes of the three MPSAs were fabricated on an RF-35 substrate (εr = 3.5, tan δ = 0.0018, h = 0.76 mm), as shown in Figure 7. The S11 characteristics of the fabricated antenna were measured using an Agilent N5230A network analyzer, and photographs of the experiment setup are shown in Figure  8. Five different standard dielectric MUTs from Taconic, Inc., were tested for the sensitivity comparison. The MUTs had relative permittivity ranging from 2.17 to 10.2, and their relative permittivity, loss tangent, and thickness from the data sheet in [39] are summarized in Table 2.
First, the S11 characteristics of the three MPSAs were simulated with the five MUTs in Table 2, as shown in Figure 9. The length of the MUTs was slightly reduced to 75 mm because there is a protruding part in the SMA connector for soldering the input port of the fabricated antenna, as shown in Figure 8b. For the conventional MPSA, the first resonant frequency for S11 moved from 2.465 GHz for the TLY-5A MUT (ε r = 2.17) to 2.312 GHz for the RF-10 MUT (ε r = 10.2). In the RS-loaded MPSA, the first resonant frequency shifted from 2.369 GHz for TLY-5A to 1.852 GHz for RF-10, whereas it moved from 2.285 GHz for TLY-5A to 1.577 GHz for RF-10 in the proposed MLS-loaded MPSA. Table  3 shows the first resonant frequencies of the simulated S11 responses for the three MPSAs.         Figure 10 compares the simulated sensitivity for the first resonant frequencies of the three MPSAs when the five MUTs were loaded. When TLY-5A was used as the MUT, the first resonant frequency shifts ∆f r for the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs were 0.035 GHz, 0.131 GHz, and 0.215 GHz, respectively, and PRFS for each of them were 1.40%, 5.24%, and 8.60%, respectively. PRFSE values for the first resonant frequency of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, ware 3.74 and 6.14, respectively. S for the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs were 0.030 GHz, 0.112 GHz, and 0.184 GHz, respectively. SE values of the RS-loaded and the proposed MLS-loaded MPSAs, compared to the conventional MPSA, were 3.74 and 6.14. When RF-10 was used as the MUT, the first resonant frequency shifts ∆f r for the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs were 0.188 GHz, 0.648 GHz, and 0.923 GHz, respectively, while PRFS was 7.52%, 25.     Next, the S 11 characteristics of the three MPSAs were measured in order to validate the simulated sensitivity for the five MUTs described in Table 2, as shown in Figure 11. The length of the MUTs was slightly reduced to 75 mm. When unloaded, the first S 11 resonant frequencies of the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs were 2.528 GHz, 2.509 GHz, and 2.506 GHz, respectively. The respective errors compared to the simulated first resonant frequency when unloaded were 1.12%, 0.36%, and 0.24%, which might have been caused by errors in fabrication and measurement. For the conventional MPSA, the first resonant frequency of S 11 moved from 2.497 GHz for TLY-5A to 2.328 GHz for RF-10, while it moved from 2.376 GHz for TLY-5A to 1.835 GHz for RF-10 in the RS-loaded MPSA. In the proposed MLS-loaded MPSA, the first resonant frequency of S 11 moved from 2.294 GHz for TLY-5A to 1.592 GHz for RF-10. We note that the S11 magnitude levels are larger than −10 dB due to impedance mismatch for the proposed MLS-loaded MPSA, but its effect on the first resonant frequency and the sensitivity is very little. Measured first resonant frequencies of the three MPSAs are summarized in Table 4.  Figure 11. Measured S11 characteristics of the three MPSAs for the MUTs in Table 2  Next, the S11 characteristics of the three MPSAs were measured in order to validate the simulated sensitivity for the five MUTs described in Table 2, as shown in Figure 11. The length of the MUTs was slightly reduced to 75 mm. When unloaded, the first S11 resonant frequencies of the conventional, the RS-loaded, and the proposed MLS-loaded MPSAs were 2.528 GHz, 2.509 GHz, and 2.506 GHz, respectively. The respective errors compared to the simulated first resonant frequency when unloaded were 1.12%, 0.36%, and 0.24%, which might have been caused by errors in fabrication and measurement. For the conventional MPSA, the first resonant frequency of S11 moved from 2.497 GHz for TLY-5A to 2.328 GHz for RF-10, while it moved from 2.376 GHz for TLY-5A to 1.835 GHz for RF-10 in the RS-loaded MPSA. In the proposed MLS-loaded MPSA, the first resonant frequency of S11 moved from 2.294 GHz for TLY-5A to 1.592 GHz for RF-10. We note that the S11 magnitude levels are larger than -10 dB due to impedance mismatch for the proposed MLS-loaded MPSA, but its effect on the first resonant frequency and the sensitivity is very little. Measured first resonant frequencies of the three MPSAs are summarized in Table 4.  The measured sensitivities for the first resonant frequencies of the three MPSAs are compared in Figure 12. When TLY-5A (ε r = 2.17), RF-301 (ε r = 2.97), TRF-43 (ε r = 4.3), RF-60A (ε r = 6.15), and RF-10 (ε r = 10.2) were used as the MUT, in turn, the first resonant frequency shifts, ∆f r , of the conventional MPSA were 0.031 GHz, 0. respectively. In the proposed MLS-loaded MPSA, S values were 0.181 GHz, 0.164 GHz, 0.146 GHz, 0.132 GHz, and 0.100 GHz, respectively. SE values for the RS-loaded MPSA were 4.29, 3.92, 3.88, 3.35, and 3.37, respectively, whereas they were 6.84, 6.08, 5.81, 5.06, and 4.57, respectively, for the proposed MLS-loaded MPSA. Hence, the measured sensitivity of the proposed MLS-loaded MPSA is 4.57 to 6.84 times higher than the conventional MPSA for the five MUTs, with relative permittivity ranging from 2.17 to 10.2. We note that the measured SE value of the proposed MPSA for low permittivity MUTs such as TLY-5A, RF-301, and RF-60A increases compared to the simulated one, whereas it decreases for the other high permittivity MUTs.  In order to validate the performance of the proposed MLS-loaded MPSA, the relative permittivity of the five MUTs were extracted by using the measured PRFS and Equation (7), and the results are compared in Table 5. It is observed that the absolute value of the maximum error between the extracted and the reference relative permittivity ranges between 0.91% and 2.68%, which is within the tolerance provided by Taconic, Inc. [39]. The measured errors result from errors in fabrication and measurement.

Conclusions
In this paper, a new, high-sensitivity MPSA has been proposed by using an MLS loaded along the radiation edge of the patch for relative permittivity measurement at 2.5 GHz. The sensitivity of the proposed MLS-loaded MPSA was compared with that of a conventional MPSA without the slot, and compared with an RS-loaded MPSA by measuring the shift in the first resonant frequency of the input reflection coefficient when the planar MUT was placed above the patch. The measured sensitivity of the proposed MPSA is 4.57 to 6.84 times higher than the conventional MPSA for the five MUTs with a relative permittivity ranging from 2.17 to 10.2. The patch size of the proposed MPSA was reduced by about 44.9%, compared to the conventional MPSA, which makes it promising for a compact permittivity sensor. Good agreement between the extracted and the reference relative permittivity with a maximum error less than 2.68% was achieved, which validates the effectiveness of the proposed MPSA.
The possible applications of the proposed high-sensitivity MPSA are permittivity characterization of planar solid materials with low relative permittivity and low loss; proximity detection in automation applications; non-invasive determination of moisture content in soil, food, or liquids; and wireless sensing of biological samples or poisonous chemicals. It can also be applicable as a simultaneous communication-and-sensing unit for chipless radio frequency identification (RFID) sensor tags measuring temperature, humidity, and other characteristics, and communicating with a reader at the same time.
In future work, we will try to characterize high relative permittivity and high loss materials using the proposed MPSA and compare its sensitivity with other MPSAs.
Author Contributions: J.Y. contributed the idea, simulation, analysis, and the overall research; J.-I.L. contributed fabrications and measurements.