This is the accepted version of the journal article: Vélez Rasero, Paris; Muñoz Enano, Jonathan; Martín, Ferran. «Differential Sensing Based on Quasi-Microstrip Mode to Slot-Mode Conversion». IEEE microwave and wireless components letters, Vol. 29, Issue 10 (October 2019), p. 690-692. DOI 10.1109/LMWC.2019.2939533 This version is available at https://ddd.uab.cat/record/258247 under the terms of the license > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1 Differential Sensing based on Quasi-Microstrip- Mode to Slot-Mode Conversion Paris Vélez, Member, IEEE, Jonathan Muñoz-Enano, Student Member IEEE, and Ferran Martín, Fellow, IEEE  the cross-mode transmission coefficient of the four-port Abstract— In this paper, a differential sensor and comparator structure (provided the lines are uncoupled) [4],[5]. This based on a mode conversion detector implemented by means of a mode-conversion approach provides high sensitivity, and has slotline resonator is proposed for the first time. The sensor been applied to the measurement of the complex permittivity divides the input signal between two identical sensing elements, a of liquid samples [4], and to the measurement of electrolyte pair of microstrip lines loaded with complementary split ring concentration in extremely diluted solutions [5]. resonators (CSRRs). Such CSRRs are symmetrically etched in To reduce system complexity, the pair of lines can be the ground plane at both sides of the slotline. Therefore, by arranged in a splitter/combiner configuration [6],[7], resulting truncating symmetry, e.g., by loading one of the CSRRs with the in a two-port based differential sensor. In this case, the output reference (REF) sample and the other one with the sample under variable is the frequency and/or the amplitude difference test (SUT), the slot mode is generated in the slotline resonator. between the pair of notches that are generated when the The signal level associated to this mode, which can be inferred by resonant elements of the lines are asymmetrically loaded. means of a slotline/microstrip transition, depends on the level of asymmetry generated by CSRR loading. Consequently, the Similar sensors, i.e., also based on frequency splitting, can be proposed structure is useful for differential sensing, where a implemented by cascading a pair of identical resonators in a single input and output port suffices to excite the sensor and to single line [8]. Although these frequency splitting sensors are measure the output variable, the magnitude of the transmission based on two-port structures, their sensitivity and resolution is coefficient. limited as compared to mode-conversion sensors. Index Terms — Microwave sensor, differential sensor, In this paper we report a two-port differential sensor for complementary split ring resonator (CSRR), mode conversion, dielectric characterization, where mode conversion is the microstrip line, slotline. sensing principle. The resolution and sensitivity of these sensors is superior to those of frequency splitting sensors, I. INTRODUCTION whilst measuring the differential dielectric constant does not P URE DIFFERENTIAL sensors are implemented by means of require a four-port measurement. a pair of identical elements, sensitive to the variable under measurement (measurand) [1]-[5]. One of such elements II. SENSOR PRINCIPLE, DESIGN AND SIMULATION provides an output signal determined by the reference The layout of the proposed sensor is depicted in Fig. 1. It measurand, whereas the output signal of the other sensing consists of a pair of identical CSRR-loaded microstrip line element is given by the measurand under test, and the segments connected through a T-junction to the input access differential output signal is merely the difference between the line. The CSRRs are symmetrically etched with regard to the individual signals generated by each sensor element. For axial plane of a slotline, implemented in the ground plane of differential permittivity sensing using microwaves, a pair of the microstrip line, as shown in Fig. 1. According to this transmission lines loaded with identical resonant elements, configuration, if the CSRRs are symmetrically loaded, the very sensitive to the properties of the surrounding medium, is symmetry plane is a magnetic wall, and the (odd) slot mode typically used [3],[5]. In these sensors, the input variable is the cannot be generated. Under these conditions, the injected input differential permittivity between a reference (REF) sample and power is reflected back to the source and partially radiated, a sample under test (SUT), whereas the output variable due to the effects of the CSRRs. By contrast, if one of the depends on the specific sensor configuration. CSRRs is loaded with a REF sample and the other one with If the pair of sensing lines are independently excited, the the SUT sample, the difference in the resonance frequencies of difference in the transmission coefficient measured at both both CSRRs causes an asymmetry able to excite the slot mode lines is typically the considered output variable. In practice, in the slotline. Since mode conversion depends on the level of such differential output variable can be obtained by measuring asymmetry, the proposed structure is useful for differential sensing. To collect the mode conversion level, a This work was supported by MINECO-Spain (project TEC2016-75650-R), slotline/microstrip transition is added to the structure, as by Generalitat de Catalunya (project 2017SGR-1159), by Institució Catalana depicted in Fig. 1 [9]. Thus, the output variable in the de Recerca i Estudis Avançats (who awarded Ferran Martín), and by FEDER funds. J. Muñoz-Enano acknowledges Secreteraria d’Universitats i Recerca proposed two-port differential sensing structure is the (Gen.Cat.) and European Social Fund for the FI grant. P. Vélez acknowledges transmission coefficient between the input and the output port. the Juan de la Cierva Program for supporting him (Project IJCI-2017-31339). The CSRRs have been designed in order to resonate at P. Vélez, J. Muñoz-Enano, and F. Martín are with GEMMA/CIMITEC (Departament d’Enginyeria Electrònica), Universitat Autònoma de Barcelona. roughly f0 = 1.5 GHz, when they are loaded with the REF 08193 BELLATERRA (Barcelona), Spain. E-mail: Ferran.Martin@uab.es. sample (in the first part of this work, the reference sample is a > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2 slab of Rogers RO4003C substrate with dielectric constant r = III. EXPERIMENTAL VALIDATION 3.55, thickness h = 1.5 mm and loss tangent tan = 0.0030). The photograph of the fabricated sensor is depicted in Fig. 3 The resulting dimensions are indicated in the caption of Fig. 1. (the LPKF H100 drilling machine has been used for sensor The considered substrate of the sensor is the Rogers RO3010 fabrication). The first experimental validation of the sensor substrate with dielectric constant r = 10.2, thickness concerns its functionality as comparator, able to detect tiny h = 1.27 mm and loss tangent tan = 0.0027. The input and differences (or defects) in a SUT, as compared to a REF output access microstrip lines, as well as the pair of lines sample. For that purpose, we have modified the REF sample between the T-junction and the CSRRs, are 50- lines (with by drilling sparse hole arrays across the substrate. The width of 1.12 mm). The microstrip line sections on top of the responses of the sensing structure are shown in Fig. 4, where CSRRs are wider in order to enhance the coupling with the the different SUTs are also depicted. It is remarkable that the CSRRs. Concerning the slot line (short-circuited at the device is able to detect the presence of a single hole, and the extremes), its width has been set to 0.4 mm, corresponding to magnitude of the transmission coefficient severely increases as a characteristic impedance of roughly 50 . Concerning the the density of holes increases. Thus, sensor resolution is good, slot length, it corresponds to one wavelength ( ) at f0, so that and sensor sensitivity with the density of holes is high. the slot resonates at such frequency. The slot-line/microstrip (a) (b) transition has been located at /4 from the right-hand side extreme of the slot, since at this position the electric field in the slot is maximized, and this favors mode conversion between the slot mode and the quasi-microstrip mode [9]. Fig. 3. Photograph of the fabricated sensor. (a) Top view; (b) bottom view Similarly, the imaginary line connecting the centers of the with REF and SUT samples attached to the substrate by means of Teflon screws. CSRRs (transverse to the slotline resonator axis) is situated at -15 /4 from the left-hand side extreme of the slotline resonator, provided this is the optimum distance for the generation of the -20 |S21|(dB) slot mode from an asymmetry in the CSRR pair. -25 LR P1 lout WR -30 ws ss 1.3 1.4 1.5 1.6 Frequency (GHz) 1.7 1.8 P1 d wc Fig. 4. Measured transmission coefficient of the sensor loaded with different ls c SUT samples consisting of the REF sample with different densities of holes ld drilled across the substrate. P2 Fig. 1. Topology of the designed sensor. Dimensions are (in mm): LR = 14.64, The sensor is also useful to estimate the dielectric constant WR = 6.50, c = d = 0.80, wc = 1.68, ws = Ss = 0.40, lout = 23.46, ls = 107.60, ld = 26.32. The ground plane where the CSRRs and the slotline resonator are of the SUT sample. It has been verified from electromagnetic simulations (not shown) that the dependence of the output P etched is depicted in grey color. -15 variable of the sensor (the maximum value of the transmission -20 -15 r(SUT)=3.55,r=0 2 coefficient magnitude) with the loss tangent of the SUT is r(SUT)=3.6565,r=0.1065 -20 r(SUT)=3.763,r=0.213 negligible. For that reason, it is possible to infer the dielectric |S21|(dB) -25 -25 r(SUT)=3.8695,r=0.3195 constant of the SUT sample from the measurement of a single -30 r(SUT)=3.976,r=0.426 -30 r(SUT)=3.4435,r=0.1065 output variable. For this purpose, air is the considered REF -35 -35 r(SUT)=3.337,r=0.213 material, provided any dielectric material exhibits a larger -40 1.3 1.4 1.5 1.6 r(SUT)=3.2305,r=0.3195 dielectric constant. Fig. 5(a) shows the measured responses of -40 r(SUT)=3.124,r=0.426 1.3 1.4 Frequency (GHz) 1.5 1.6 the sensor for different SUT samples (Arlon, FR4 and RO3010 Fig. 2. Transmission Frequencycoefficient (GHz) of the structure of Fig. 1 corresponding to slabs with dielectric constants 2.43, 4.5 and 1.02 respectively), several differential loadings of the CSRRs, as indicated in the figure. whereas Fig. 5(b) depicts the measured differential maximum transmission coefficient (∆|S21|max = |S21|max,REF  |S21|max,SUT), To verify the functionality of the proposed structure as with the differential dielectric constants (∆εr = εr,REF  εr,SUT). differential sensor, we have carried out simulations by From such data points, a calibration curve has been inferred considering the above-cited REF material, and a set of (by means of a curve fitting tool software), and such curve can hypothetical SUTs obtained from the REF sample by be used to determine the differential dielectric constant from perturbing its dielectric constant, maintaining the loss tangent the measurement of the differential maximum value of the unaltered. The results, inferred from the ANSYS HFSS transmission coefficient. To demonstrate the potential of the commercial simulator, are depicted in Fig. 2. It can be seen structure as differential sensor, able to provide the dielectric that the transmission coefficient experiences significant constant of the SUT sample in real time, we have considered variations when the dielectric constant is perturbed. Note that several SUTs samples. Particularly, a piece of Rogers increasing/decreasing the dielectric constant of the SUT with RO4003C (r,SUT = 3.55) substrate and two slabs of 3D-printed regard to the REF sample a certain level has the effect of materials (fabricated by means of the Ultimaker 3 Ext. model shifting down/up the frequency of maximum transmission, 3D-printed). Such 3D-printed materials are PLA (Polylactic keeping such transmission level unaltered. Thus, it is possible Acid) and MT-COPPER with nominal dielectric constants of 3 to determine the sign of the differential dielectric constant. > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3 and 7.6 respectively (those values have been obtained using a good accuracy, but the approach needs smd components (i.e., Agilent 85072A resonant cavity). Note that the thickness of the 50 Ω resistors). considered samples is roughly the same (1.5 mm), in order to V. CONCLUSIONS avoid cross sensitivities against thickness effects. The responses with such SUT samples are also depicted in Fig. 5. In conclusion, a differential sensor and comparator based Considering the maximum transmission coefficient of each on a quasi-microstrip-mode to slot-mode conversion detector sample, the extracted differential dielectric constant for has been reported. The sensitive part of the sensor consists of RO4003C, PLA and MT-COPPER are Δεr = 2.42, Δεr = 2.10, a pair of CSRR-loaded lines fed by a common port, and placed and Δεr = 6.02 respectively. With these results, and symmetrically at both sides of a slotline resonator. By considering Air as the reference sample, the measured differential loading of the CSRRs, symmetry is truncated and dielectric constant of each sample is 3.42, 3.10 and 7.02, the slotline resonator is excited, thereby generating an output respectively (i.e., the sensor accuracy is better than 92.4%). signal at the output (microstrip) port of the structure, thanks to 0 RO3010 10 the coupling between the slot mode and the quasi-microstrip (a) (b) r=0.00783|S21|+0.04331|S21|2 -5 MT-COPPER 8 -0.00451|S |3+0.000140623|S RO3010 4 mode. It has been demonstrated that the structure is useful to | 21 21 detect small differences between samples (comparator |S21|(dB) -10 FR4 Measurement data 6 MT-COPPER PLA calibration curve functionality), and to measure the dielectric constant of r -15 RO4003C (nominal value) 4 -20 ARLON PLA FR4 (nominal value) RO4003C dielectric slabs (samples under test SUTs). 2 -25 Air Arlon (nominal value) 0 Air -30 0 5 10 15 20 25 REFERENCES 1.3 1.4 1.5 1.6 1.7 Frequency (GHz) |S21|(dB) [1] C. Damm, et al., “Artificial transmission lines for high sensitive Fig. 5. (a) Measured transmission coefficient of the sensor by considering air microwave sensors,” IEEE Sensors Conf., Christchurch, New Zealand, as REF material and different SUT samples; (b) dependence of the differential 2009, pp. 755–758. dielectric constant with the differential transmission coefficient and [2] F.J. Ferrández-Pastor, J.M. García-Chamizo and M. Nieto-Hidalgo, calibration curve. “Electromagnetic differential measuring method: application in microstrip sensors developing”, Sensors, vol. 17, p. 1650, 2017. [3] A. Ebrahimi, J. Scott and K. Ghorbani, "Differential sensors using IV. DISCUSSION AND COMPARISON TO OTHER SENSORS microstrip lines loaded with two split ring resonators," IEEE Sensors J., Most microwave sensors for dielectric constant vol. 18, pp. 5786-5793, Jul. 2018. [4] P. Vélez, et al., “Highly-sensitive microwave sensors based on open measurements are based on frequency variation [10]-[16] or complementary split ring resonators (OCSRRs) for dielectric frequency splitting [6]-[8]. Thus, comparing the sensor characterization and solute concentration measurements in liquids”, proposed in this work with those frequency-based sensors is IEEE Access, vol. 6, pp. 48324-48338, Dec. 2018. [5] P. Vélez, et al., ”Split ring resonator (SRR) based microwave fluidic not straightforward. Nevertheless, the reported sensor exhibits sensor for electrolyte concentration measurements”, IEEE Sensors J., excellent resolution, as derived from the fact that minuscule vol. 19, no. 7, pp. 2562-2569, Apr. 2019. perturbations in material composition are detected (see Fig. 4, [6] L. Su, et al., “Splitter/combiner microstrip sections loaded with pairs of complementary split ring resonators (CSRRs): modeling and where the presence of a single hole in the REF material is optimization for differential sensing applications,” IEEE Trans. Microw. perfectly detected). In [4], the reported differential sensor is Theory Techn., vol. 64, pp. 4362-4370, Dec. 2016. based on the variation of the cross mode transmission [7] P. Vélez, et al., “Microwave microfluidic sensor based on a microstrip coefficient, and the measured (maximum) sensitivity with the splitter/combiner configuration and split ring resonators (SRR) for dielectric characterization of liquids”, IEEE Sensors J., vol. 17, pp. differential dielectric constant is found to be roughly 0.43 dB, 6589-6598, Oct. 2017. i.e., not as good as the one inferred in this paper (8.10 dB). In [8] J. Naqui, et al., “Transmission Lines Loaded with Pairs of Stepped [4], the effects of the dielectric constant and loss factor of the Impedance Resonators: Modeling and Application to Differential Permittivity Measurements”, IEEE Trans. Microw Theory Techn., vol. SUT on the output variables (magnitude and frequency of the 64, no. 11, pp. 3864-3877, Nov. 2016. maximum cross mode insertion loss) are not decoupled. By [9] X. Guo, et al., “Wideband differential bandpass filters on multimode contrast, the proposed sensor is insensitive to the loss tangent slotline resonator with intrinsic common-mode suppression,” IEEE Trans. Microw Theory Techn., vol. 63, pp. 1587-1594, May. 2015. of the SUT, and, therefore, the dielectric constant can be [10] M. S. Boybay and O. M. Ramahi, “Material characterization using inferred from a simple single-ended measurement, involving complementary split-ring resonators,” IEEE Trans. Instrum. Meas., vol. one output variable. 61, no. 11, pp. 3039−3046, Nov. 2012. [11] C.-S. Lee and C.-L. Yang, “Complementary split-ring resonators for TABLE I measuring dielectric constants and loss tangents,” IEEE Microw. COMPARISON OF VARIOUS SENSORS WITH GOOD ACCURACY Wireless Compon. Lett., vol. 24, no. 8, pp. 563−565, Aug. 2014. Ref. f0 (GHz) Accuracy (%) Technology Diff. Meas. [12] C.-L. Yang, et. al., “Noncontact measurement of complex permittivity [10] 1.286 96.46 / 97.36 Planar No and thickness by using planar resonators,” IEEE Trans. Microw. Theory [14] 2.19 95.00 / 98.6 / 99.13/ Planar No Techn., vol. 64, no.1, pp. 247−257, Jan. 2016. 99.77 [13] L. Su, et al., “Analytical method to estimate the complex permittivity of [16] 2.41 99.09 / 97.87 / 98.13 Planar+ smd Yes oil samples”, Sensors, vol. 18, p. 984, 2018. T. W. 1.5 96.6 / 95.4 / 92.4 Planar Yes [14] R. A. Alahnomi, et al., “High-q sensor based on symmetrical split ring resonator with spurlines for solids material detection,” IEEE Sensors J., Table I compares some dielectric constant sensors exhibiting vol. 17, no. 9, pp. 2766–2775, 2017 good accuracy. The sensors reported in [10],[14] exhibit [15] T. Haq, et al., “Complementary metamaterial sensor for nondestructive comparable accuracy to ours, but they do not operate evaluation of dielectric substrates,” Sensors, vol. 19, p. 2100, 2019. [16] A. Ebrahimi, et al., "Transmission lines terminated with LC resonators differentially. The differential sensor of [16] exhibits very for differential permittivity sensing", IEEE Microwave and Wireless Components Letters, vol. 28, no. 12, pp. 1149-1151, Dec. 2018.