sensors - MDPI

3 downloads 0 Views 5MB Size Report
Jun 18, 2018 - consideration of two open edges as magnetic walls and one electric wall along the hypotenuse. In addition to TE modes, transverse-magnetic ...

sensors Article

TM02 Quarter-Mode Substrate-Integrated Waveguide Resonator for Dual Detection of Chemicals Ahmed Salim and Sungjoon Lim *

ID

School of Electrical and Electronics Engineering, College of Engineering, Chung-Ang University, 221, Heukseok-Dong, Dongjak-Gu, Seoul 156-756, Korea; [email protected] * Correspondence: [email protected]; Tel.: +82-2-820-5827; Fax: +82-2-812-7431 Received: 19 May 2018; Accepted: 16 June 2018; Published: 18 June 2018

 

Abstract: The detection of multiple fluids using a single chip has been attracting attention recently. A TM02 quarter-mode substrate-integrated waveguide resonator designed at 5.81 GHz on RT/duroid 6010LM with a return loss of 13 dB and an unloaded quality factor of Q ≈ 13 generates two distinct strong electric fields that can be manipulated to simultaneously detect two chemicals. Two asymmetric channels engraved in a polydimethylsiloxane sheet are loaded with analyte to produce a unique resonance frequency in each case, regardless of the dielectric constants of the liquids. Keeping in view the nature of lossy liquids such as ethanol, the initial structure and channels are optimized to ensure a reasonable return loss even in the case of loading lossy liquids. After loading the empty channels, Q is evaluated as 43. Ethanol (E) and deionized water (DI) are simultaneously loaded to demonstrate the detection of all possible combinations: [Air, Air], [E, DI], [DI, E], [E, E], and [DI, DI]. The proposed structure is miniaturized while exhibiting a performance comparable to that of existing multichannel microwave chemical sensors. dual detection; TM02 -mode quarter-mode substrate-integrated waveguide; Keywords: microwave sensor; microfluidic channel

1. Introduction The monitoring of several parameters at different processing steps is common in chemical industries and pharmaceutical plants and in quality control for the food industry [1]. Multiple sensors or a sensor array can be employed in these scenarios; however, these approaches have a large footprint, a complex design, and/or expensive fabrication for mass production. A radiofrequency (RF) sensor chip realized from components and monolithic devices can be utilized as a multiple sensing device; however, its high power consumption could limit its widespread adoption. As an alternative method, simultaneous detection of multiple fluids using a single-chip sensor is the main objective of this study. Radio frequency (RF) technology-based sensors are generally noninvasive, small, inexpensive, and easily fabricated compared with nonelectromagnetic sensors. Moreover, they are noncontact and operate in the ambient environment. However, their sensitivity (lower limit of detection) is significantly lower than that of nonelectromagnetic sensors. For instance, in [2], a microtoroid-based label-free optical resonator (which works on total internal reflection) exhibited an excellent limit of detection on the order of a single protein molecule (2.5-nm radius). In addition, RF sensors lack selectivity, which is a serious drawback from a practical viewpoint. Trends in RF sensors include enhancing the performance, miniaturization, and adding functionality, such as the capability to detect multiple chemicals. Additionally, hybrid sensors (RF sensors with additional coatings) have been utilized to obtain RF sensors with selectivity. Rectangular waveguide resonators render electric-field (E-field) energy confined in a localized area and reduce radiation losses and effects that arise from a high resistance [3]. Substrate-integrated Sensors 2018, 18, 1964; doi:10.3390/s18061964

www.mdpi.com/journal/sensors

Sensors 2018, 18, 1964

2 of 18

waveguide (SIW) technology exhibits advantages of rectangular waveguides (such as low radiation losses, low leakage losses, and high Q) and planar circuit boards (such as a planar structure, easy fabrication, small size, and lightweight structures) [4–6]. A strong E-field (energy) centered at the SIW cavity can be utilized for various sensing applications. A high quality-factor (Q ≈ 300) SIW-based microwave humidity sensor has been proposed [7]. To realize a functionalized region on the SIW, air holes were introduced on the strongest E-field. Without using any additional coating layers, sensitivity was enhanced owing to a larger functionalized area. An SIW resonator working in transverse-electric (TE) modes specifically TE101 and TE102 was proposed as a gas detection sensor [8]. Tin oxide (SnO2 ) powder, being reactive toward hydrogen gas (analyte), was poured to functionalize a specific area (encompassing the strongest E-field). To investigate the effect of material-dependent properties on the sensitivity, several analyses were conducted, such as dimensions of the functionalized area, substrate permittivity, and grain size of the SnO2 powder on the sensitivity of the device. The transverse-electric (TE) modes of rectangular metallic waveguides also exist in SIWs because the propagation characteristics of the SIW are similar to those of the metallic waveguide, provided that the metallic vias of the SIW are closely spaced and the radiation leakage is neglected [9]. In the following paragraph, the evolution of the quarter-mode substrate-integrated waveguide (QMSIW) is detailed. Several attempts have been made to reduce the size of SIWs while maintaining or even enhancing the performance. Various RF applications have been realized, such as half-mode SIWs (HMSIWs) [10,11], quarter-mode SIWs (QMSIWs) [12,13], eight-mode SIWs (EMSIWs) [14,15], sixteen-mode SIWs [16], and ridged quarter-mode SIWs [17]. According to image theory, in-phase fields exist on opposite sides; thus, the symmetric plane of an SIW is cut into two halves to obtain an HMSIW [12]. The HMSIW has almost the same E-field distribution as the full SIW, while its size is halved. Owing to the fringing fields on the edges of the HMSIW, the equivalent size of the HMSIW is smaller than the half of that of the corresponding full SIW [18]. Further bisection of an HMSIW is performed to obtain the quarter-mode SIW (QMSIW). In a QMSIW, the E fields are further reduced to some extent; however, the QMSIW is adequate for sensing and tuning applications [17–19]. Being a quadrant of an SIW, the size of a QMSIW is one-fourth that of a full SIW. In [18], a QMSIW was excited in the TE101 and TE202 modes to achieve linearly polarized and circularly polarized radiation, respectively. In [19], a triple-mode filter was realized using an isosceles right-angle triangular waveguide with consideration of two open edges as magnetic walls and one electric wall along the hypotenuse. In addition to TE modes, transverse-magnetic (TM) modes are supported in a triangular QMSIW structure [19]. In [20], a novel flow sensor was proposed to measure the fluidic flow. A microfluidic channel covered with a polydimethylsiloxane (PDMS)-based membrane served as the microfluidic device which was integrated with a half-wavelength microwave resonator. The highest sensitivity recorded with this noncontact and nonintrusive flow sensor was 0.5 µL/min for a thin membrane which was 3 mm in diameter and 100 µm thick. Metamaterial (MM) unit cells are arranged to realize periodic structures [21,22] that simultaneously exhibit negative permittivity and negative permeability [23–25]. MM-inspired structures provide a higher quality factor (Q) and are smaller than SIW-based cavity resonators, half-wavelength resonators, and so forth. The geometry, topology, and way in which they are configured, using either an electric or a magnetic field, determine their response and operating-frequency range [26,27]. MM topologies, for example, split-ring resonators (SRRs) and complementary SRRs, are widely utilized to realize diverse RF components [28,29]. Recently, SRR-based chemical sensors have been proposed for detecting two or more chemicals. For instance, in [30], four unique resonance frequencies were tuned using 5 µL of ethanol in an experimental demonstration. In [31], a microfluidic dual-channel resonator consisting of three open-SRRs with unequal dimensions was reported. The independent tuning of one resonance was demonstrated. In [32], two SRRs coupled with a microstrip line were proposed as a dual-detection chemical sensor. This sensor is based on the frequency-splitting phenomenon and requires two asymmetric dielectric chemicals to be loaded. If the same dielectric chemical was loaded in both channels, a frequency

Sensors 2018, 18, 1964

3 of 18

shift was not observed. In [33], a dual-mode resonator was proposed as a dual sensor. In one mode, the resonance-shift principle is based on the capacitive effect, while in the other mode, the properties of a stub (i.e., the change of the characteristic impedance and the electrical length) are used. In Section 5, we compare the performance of our proposed sensor and the aforementioned dual/multiple detection RF sensors. The TM02 -mode QMSIW design used in this study is inspired by [18,19], and we applied it to extend our previous work [34], in which a TE20 -mode full SIW was proposed as a dual-detection microwave chemical sensor. The main contribution of the present study is miniaturization together with simultaneous dual detection using a microwave sensor. The size is reduced mainly by utilizing a QMSIW resonator and a substrate with a high dielectric constant. Unlike our previous dual-detection microwave sensor [34], in which polydimethylsiloxane (PDMS) was inserted between two main substrates to realize a sandwich-like structure, only a single main substrate (Rogers RT/duroid) is utilized in the present study. The details and analysis are presented in Section 2. Herein, a TM02 -mode QMSIW resonator that can simultaneously detect two analytes using a single-chip sensor is proposed. Two asymmetric microfluidic channels have been designed to be loaded on two distinct E-field regions to perturb the effective permittivity of the localized regions in the substrate. All possible combinations of ethanol and deionized (DI) water are loaded in the channels, and the distinct resonance frequency in each case is reported. The design guidelines, fabrication, and measurements are illustrated. In addition, the salient features of the proposed sensor are compared with those of existing multiple-detection sensors. 2. Sensor Design 2.1. Theory The dielectric constant (relative permittivity), which indicates an ability associated with dielectric materials, tends to measure how easy or difficult it is to polarize a material upon excitation of the external E-field and the consequent storage of energy. Rogers Corporation Inc. (Chandler, AZ, USA) provides specially designed microwave low-loss substrates with highly stable dielectric constants, such as 5880, 5870, 3010, and 6010LM [35,36]. Substrate materials having low dielectric constants (εr ) have been extensively utilized in various RF designs. The choice of the feeding technique, design approach/technology, and substrate material plays a key role in the performance and size of the design. Substrates having high dielectric constants (εr ≈ 10) are known to facilitate miniaturization [37]. To find a suitable substrate, a design with the same dimensions of the triangular patch are realized on two different substrates: namely, RT/duroid 5880 (εr = 2.2, thickness 0.51 mm) and RT/duroid 6010LM (εr = 10.7, thickness hs =1.27 mm). The resonance frequency is observed at 12.91 and 5.82 GHz, and the results for the return loss obtained from these simulations are shown in Figure 1. The 6010LM substrate is an obvious choice for developing a compact TM02 -mode QMSIW resonator without compromising on performance; therefore, it is chosen as the main substrate in this study.

Sensors 2018, 18, 1964

4 of 18

Sensors 2018, 18, x

4 of 18

Figure 1. TM02-mode QMSIW resonator realized on two substrates having significantly different Figure 1. TM02 -mode QMSIW resonator realized on two substrates having significantly different dielectric constants; the corresponding return loss (S11) and resonance frequencies are shown. dielectric constants; the corresponding return loss (S11) and resonance frequencies are shown. RT/duroid 6010LM (εr = 10.7) exhibits a significant downshift in the resonance frequency, which RT/duroid 6010LM (εr = 10.7) exhibits a significant downshift in the resonance frequency, which ensures ensures a compact design compared with the design realized on RT/duroid 5880 (εr = 2.2). The TM02a compact design compared with the design realized on RT/duroid 5880 (εr = 2.2). The TM02 -mode mode in each case is verified after analyzing the E-field magnitude distribution, which is not shown in each case is verified after analyzing the E-field magnitude distribution, which is not shown here here for brevity. for brevity.

If the height (thickness of substrate used, in our case hs = 1.27 mm) of an SIW cavity resonator is If the height (thickness used, case hs = 1.27 frequency mm) of an of SIW resonator is far smaller than its length (LofSIWsubstrate ) and width (WinSIWour ), the resonance thecavity resonator can be far smaller than its length (LSIW ) and width (WSIW ), the resonance frequency of the resonator can be defined as defined as s     nπ 2 2 1 mπ 22     m + n (1) f mn0 = √1 (1) f mn 0  2π µε WSIW    LSIW

2 

  respectively,  where µ and ε represent the permittivity and permeability, of the dielectric material and m and n are the integer-mode indices. According to Equation (1), TE the lowest resonance where µand ε represent the permittivity and permeability, respectively, of100 theisdielectric material and frequency [13]the andinteger-mode TE101 is the dominant resonant mode [38]. The (1), desired frequency can be m and n are indices. According to Equation TE100resonant is the lowest resonance tuned by adjusting length width of the SIWmode cavity resonator and the effective permittivity of frequency [13] and the TE101 is theand dominant resonant [38]. The desired resonant frequency can be the dielectric material. Because a QMSIW almost E-field distribution tuned by adjusting the length and width ofmaintains/preserves the SIW cavity resonator andthe thesame effective permittivity of as an SIW, Equation (1) can be applied to determine the resonance frequency in a specific mode of the dielectric material. Because a QMSIW maintains/preserves almost the same E-field distribution aasQMSIW. an SIW, Equation (1) can be applied to determines the resonance frequency in a specific mode of a  2 1 2π QMSIW. (2) f 02 = f 20 = √ 2π µε WQMSIW WSIW

LSIW

2

1 (2) is modified 2  to include the fringing effects For a TM02 -mode QMSIW resonator, Equation (2) f  f    02 20 arising from the open-ended edges of the triangular W 2 QMSIW: 

QMSIW



1 2π f 02 = Equation (3) √ For a TM02-mode QMSIW resonator, (2) is modified to include the fringing effects 2π µε WQMSIW + ∆W arising from the open-ended edges of the triangular QMSIW: where WQMSIW is the physical width of the 1 QMSIW patch 2 and ∆W is the extended width due to f  fringing fields. Therefore, the dimensions of this QMSIW cavity resonator are smaller than those of (3)a 02 2  WQMSIW  W quarter of an SIW.

where WQMSIW is the physical width of the QMSIW patch and ΔW is the extended width due to fringing fields. Therefore, the dimensions of this QMSIW cavity resonator are smaller than those of a quarter of an SIW.

Sensors 2018, 18, x 1964

5 of 18

2.2. Design of Dual-Detection Chemical Sensor 2.2. Design of Dual-Detection Chemical Sensor In this subsection, we describe the design process for the TM02-mode triangular-patch QMSIW. In this subsection, describepatch the design processare forrealized the TMon 02 -mode The conductive QMSIWwe triangular and ground the toptriangular-patch and bottom of aQMSIW. Rogers The conductive QMSIW triangular patch and ground are realized on the top and bottom ofsubstrate a Rogers RT/duroid 6010LM substrate (εr = 10.7, tan Δ = 0.0023, hs = 1.27 mm). The dimensions of the RT/duroid 10.7, tan ∆ =of0.0023, hs = 1.27 mm).QMSIW The dimensions ofare thechosen substrate r =dimensions are Ls × Ws =6010LM 30 mm substrate × 30 mm. (ε The the right-triangular resonator as are L × W = 30 mm × 30 mm. The dimensions of the right-triangular QMSIW resonator are chosen s s W ×L = 17 mm × 15 mm to resonate at 5.81 GHz. There is no reason to choose this particular frequency as W ×that L =testing 17 mmand × 15 mm to resonate at 5.81 GHz. microwave There is no sensors reason to choose particular except comparison with state-of-the-art show it to this be reasonable. frequency except that are testing andalong comparison with state-of-the-art microwave it toand be Metallic vias (copper) infused the hypotenuse of the triangular patch tosensors connectshow the top reasonable. Metallic vias (copper) are infused along the hypotenuse of the triangular patch to connect bottom grounds for realizing an electric wall along the hypotenuse and two magnetic sidewalls along the top and bottom grounds fortriangular realizing an electricresonator. wall alongConsequently, the hypotenuse and two magnetic the open-ended edges of the QMSIW the E-field remains sidewalls along the open-ended edges of the triangular QMSIW resonator. Consequently, the E-field preserved inside the triangular patch, which can be used for sensing. The performance of the QMSIW remains preserved inside the triangular patch, which can be used for sensing. The performance is improved by minimizing the leakage/radiation losses, such that D < λg/5 and P < 2D, where D, of P, the QMSIW is via improved by the minimizing thethe leakage/radiation losses, such that D < λg /5 and P < 2D, and λg are the diameter, pitch, and guided wavelength, respectively [39,40]. The values of where D,are P, determined and λg are the via0.6 diameter, the pitch, and the A guided wavelength, [39,40]. D and P to be and 1.1 mm, respectively. 50 Ω microstrip linerespectively is commonly used The values of D and P are determined to be 0.6 and 1.1 mm, respectively. A 50 Ω microstrip line to excite the microwave resonators as a preferred feeding technique. However, once the dielectric is commonly used to excite microwave resonators a preferred feeding technique. However, constant and thickness of thethe substrate are chosen, the as width of the microstrip line is almost fixed, once the dielectric constant and thickness of the substrate are chosen, the width of the microstrip with a narrow margin of variation. The dimensions (length and width) of the microstrip line areline Lm is × almost fixed, with a narrow margin of variation. The dimensions (length and width) of the microstrip Wm = 10.5 mm × 1.1 mm. The microstrip line (50 Ω) coupled with the triangular patch realized on line are Lfailed = 10.5 mm × 1.1 mm. The microstrip line (50 Ω) coupled the with the triangular patch m × Wto m produce 6010LM reasonable impedance matching. Alternatively, coupling-gap feeding realized on 6010LM failed to produce reasonable impedance matching. Alternatively, the coupling-gap technique is considered. In order to find the required coupling, various gaps (G) between the feeding technique is considered. In order findinvestigated, the required as coupling, gaps between triangular patch and the microstrip linetoare shown various in Figure 2. (G) Three casesthe of triangular patch and the microstrip line are investigated, as shown in Figure 2. Three cases of coupling coupling gap (G = 0.1 mm, 0.2 mm, and 0.3 mm) are considered. Variation in notch frequency of 40 gap (Gand = 0.1 and 0.3 mm)Gare considered. Variation notch of 400.2 MHz and MHz 10 mm, MHz0.2 aremm, observed when changes from 0. 1 mm toin0.2 mm frequency and then from mm to 10 are observedGwhen changes from 0. to 1 mm 0.2 mm and then from 0.2after mmthe to 0.3 mm, 0.3 MHz mm, respectively. = 0.2 G mm is determined yieldtocritical coupling; however, loading respectively. G = (discussed 0.2 mm is determined yield criticalthe coupling; however, aftergap the should loadingbe of the of the channels in the nexttosubsection), optimum coupling rechannels (discussed in the next subsection), the optimum coupling gap should be re-investigated. investigated.

Figure 2. Return loss and notch frequency variation for various coupling gaps (G) between the Figure 2. Return loss and notch frequency variation for various coupling gaps (G) between the microstrip line and the TM02-mode QMSIW triangular patch. The coupling gap of G = 0.2 mm yields microstrip line and the TM02 -mode QMSIW triangular patch. The coupling gap of G = 0.2 mm yields the best impedance matching before the loading of the microfluidic channels. The largest notch the best impedance matching before the loading of the microfluidic channels. The largest notch frequency variation is 40 MHz while G changes from 0.1 mm to 0.2 mm. frequency variation is 40 MHz while G changes from 0.1 mm to 0.2 mm.

The physical size of the proposed TM02-mode QMSIW resonator, including the microstrip line, is 25.6 mm × 17 mm. Its electrical size is 1.62λg × 1.07λg at 5.81 GHz. The proposed TM02-mode QMSIW resonator—without the loading of microfluidic channels—is shown in Figure 3.

Sensors 2018, 18, 1964

6 of 18

The physical size of the proposed TM02 -mode QMSIW resonator, including the microstrip line, is 25.6 mm × 17 mm. Its electrical size is 1.62λg × 1.07λg at 5.81 GHz. The proposed TM02 -mode QMSIW resonator—without the loading of microfluidic channels—is shown in Figure 3. Sensors 2018, 18, x 6 of 18

(a)

(b) Figure 3. Geometry of TM the TM 02-mode QMSIW resonator proposed as a dual-detection chemical Figure 3. Geometry of the 02 -mode QMSIW resonator proposed as a dual-detection chemical sensor: sensor: (a) top view. Ls and Ws represent substrate’s (length and width), Lm and Wm represent (a) top view. Ls and Ws represent substrate’s (length and width), Lm and Wm represent microstrip line’s microstrip line’s (length and width), OA, AB and OB represent dimensions of triangular patch; (b) (length and width), OA, AB and OB represent dimensions of triangular patch; (b) cross-sectional view. cross-sectional view.

2.3.2.3. Design of Asymmetric Microfluidic Design of Asymmetric MicrofluidicChannels Channels To change thethe effective permittivity material,microfluidic microfluidic channels loaded To change effective permittivityof ofthe the dielectric dielectric material, channels areare loaded in the strongest E-field regions; sensitivitycan canbebeexpected. expected.TheThe E-field in the strongest E-field regions;consequently, consequently, aa higher higher sensitivity E-field magnitude of the TM 02 -mode QMSIW resonator is shown in Figure 4. Two distinct E-field regions magnitude of the TM02 -mode QMSIW resonator is shown in Figure 4. Two distinct E-field regions can can be manipulated two microfluidic channels. In our previousstudy, study,two twosymmetric symmetric channels be manipulated using using two microfluidic channels. In our previous channels of of equal fluid-carrying capacity unable to produce distinct resonance frequencies when loadedwith equal fluid-carrying capacity werewere unable to produce distinct resonance frequencies when loaded with [E, DI water] and [DI water, E] [34]. Therefore, two channels having unequal fluid-carrying [E, DI water] and [DI water, E] [34]. Therefore, two channels having unequal fluid-carrying capacities capacities must be realized. must be realized. To investigate the influence of the volume in the dual-detection process, two meander-shaped To investigate the influence of the volume in the dual-detection process, two meander-shaped asymmetric channels (Ch A and Ch B) are designed, as shown in Figure 5. Although their fluidasymmetric channels (Ch A and Ch B) are designed, as shown in Figure 5. Although their fluid-carrying carrying capacities are unequal—the volume of Ch A = 17.76 µ L and the volume of Ch B = 13.11 µ L— capacities are unequal—the volume of Ch A = 17.76 µL and the volume of Ch B = 13.11 µL—the the resonance frequency in each channel-loading case was not unique. According to the dimensions, resonance frequency inofeach case was notand unique. to the the effective volumes Ch Achannel-loading and Ch B are estimated as 8.15 10.50 µ According L, respectively. It is dimensions, expected the that effective volumes of Ch A and Ch B are estimated as 8.15 and 10.50 µL, respectively. expected the difference between the effective volumes of these channels is insufficient to generateItaisunique thatresonant the difference between the effective volumes of these channels is insufficient to generate a unique frequency in each case. resonant frequency in each case.

Sensors 2018, 18, 1964

7 of 18

Sensors 2018, 18, Sensors 2018, 18,x x

7 of 7 of 1818

(a) (a)

(b) (b)

(c) (c) Figure 4. (a) E-field magnitude TM0202-mode -modeQMSIW QMSIW resonator at the resonance Figure 4. (a) E-field magnitudeofofthe the proposed proposed TM resonator at the resonance Figure 4. (a) E-field magnitude of the proposed TM02-mode QMSIW resonator at the resonance frequency of 5.81 GHz. Two distinct regionsare arethe thebest best place load microfluidic frequency of 5.81 GHz. Two distinctstrong strong E-field E-field regions place to to load twotwo microfluidic frequency of 5.81 GHz. Two distinct strong E-field regions are the best place to load two microfluidic channels. E-field vectors (verticaldistribution) distribution) of of the proposed 0202 -mode QMSIW resonator during channels. E-field vectors (vertical proposedTM TM -mode QMSIW resonator during channels. E-field vectors (vertical of the proposed 02-mode QMSIW resonator during a positive cycle adistribution) negative half cycle are shown.TM There cleardistinction distinction between (b) a(b) positive halfhalf cycle andand (c) (c) a negative half cycle are shown. There isisaaclear between the (b) a two positive half cycle regions, and (c) aasnegative cycle areensures shown. There is a clear distinction between shown inhalf (a,b). realization two closely spaced spaced yet twothe strong strong E-fieldE-field regions, as shown in (a,b). ThisThis ensures thethe realization ofoftwo closely the two strong E-field regions, as shown in (a,b). This ensures the realization of two closely spaced yet noninterfering channels. noninterfering channels. yet noninterfering channels.

Figure 5. Simulation of two asymmetric channels, unable to produce a unique resonant frequency in each5.case, which can explained by reasoning that the effective volumes (equivalent fluidic volumein in Figure 5. Simulation two asymmetric channels, unable to a aunique resonant frequency Figure Simulation ofofbe two asymmetric channels, unable toproduce produce unique resonant frequency that overlaps with the strongest E-field) of the channels differ by a narrow margin. Although the each case, whichcan canbebeexplained explainedby by reasoning reasoning that fluidic volume each case, which that the theeffective effectivevolumes volumes(equivalent (equivalent fluidic volume channels are intentionally designed to be asymmetric, the effective permittivity is perturbed in a that overlapswith withthe thestrongest strongestE-field) E-field) of of the the channels channels differ thethe that overlaps differ by by aa narrow narrowmargin. margin.Although Although uniform way owing to the inadequate difference between the effective volume of channels: Ch A and channels are intentionally designed to be asymmetric, the effective permittivity is perturbed in a a channels are intentionally designed to be asymmetric, the effective permittivity is perturbed in Ch B. Therefore, different combinations of channel loadings—such as [E, DI], [DI, DI], and [Air, DI]— uniformway wayowing owing to to the the inadequate between thethe effective volume of channels: Ch A and uniform inadequatedifference difference between effective volume of channels: Ch A are observed to generate exactly the same resonance frequency. E and DI represent ethanol and ChCh B. Therefore, different combinations of channel loadings—such as [E, DI], [DI,DI], DI],[DI, and DI], [Air,and DI]— and B. Therefore, different combinations of channel loadings—such as [E, [Air, deionized water, respectively. are observed to generate exactly thethe same resonance frequency. E and DI DI represent ethanol andand DI]—are observed to generate exactly same resonance frequency. E and represent ethanol deionized water, respectively. deionized water, respectively.

Sensors 2018, 18, 1964

8 of 18

Sensors 2018, 18, x

8 of 18

2.4. Optimized Optimized Geometry Geometry and and Effective Effective Volume Volume of Microfluidic Channels 2.4. of Microfluidic Channels If the of both both channels channels is constant and and the the overlapped overlapped area area of one channel channel differs If the thickness thickness of is kept kept constant of one differs significantly from that of the other channel, different resonance frequencies in each case can be expected. significantly from that of the other channel, different resonance frequencies in each case can be The overlapped area represents the only part of the channel covers the E-field the QMSIW expected. The overlapped area represents the only part of thethat channel that covers theonE-field on the patch, ignoring the part of the channel that covers the 6010LM substrate owing to its noninfluential QMSIW patch, ignoring the part of the channel that covers the 6010LM substrate owing to its contribution. To overlap theTo maximum area of the strongest E-field, the channels are channels designedare as noninfluential contribution. overlap the maximum area of the strongest E-field, the inverted-U-shaped (Ch 1) and meander-shaped (Ch 2). The length PDMS sheet containing Ch 1 designed as inverted-U-shaped (Ch 1) and meander-shaped (Ch of 2).the The length of the PDMS sheet and the width of the PDMS sheet containing Ch 2 are designed to eliminate the need for alignment containing Ch 1 and the width of the PDMS sheet containing Ch 2 are designed to eliminate the need marks. Both channels—each having a depth of hc = 0.6 mm—are in a PDMS layer (hp a= PDMS 1mm). for alignment marks. Both channels—each having a depth of hengraved c = 0.6 mm—are engraved in A double-sided adhesive bonding adhesive film (εr =bonding 3.4, tan ∆film = 0.03, mm) and is considered below f = 0.05 layer (hp = 1mm). A double-sided (εr =and 3.4,htan Δ = 0.03, hf = 0.05 mm) is for the PDMS-based channels to ensure the noncontact feature of the microwave sensor as well as to considered below for the PDMS-based channels to ensure the noncontact feature of the microwave maintain of the PDMS. sensor as the wellposition as to maintain the position of the PDMS. After the microfluidic channels are are loaded loadedwith withthe thelossy lossyliquids, liquids,the theimpedance impedance matching After the microfluidic channels matching is is disrupted and the relevant design parameters (e.g., the coupling gap G) must be redesigned. disrupted and the relevant design parameters (e.g., the coupling gap G) must be redesigned. Because Becauseisethanol a lossy liquid with high lossinstead tangent, of emptyboth channels, bothfilled channels ethanol a lossyis liquid with a high lossa tangent, of instead empty channels, channels with filled with ethanol are subjected to impedance matching. A parametric analysis is conducted to ethanol are subjected to impedance matching. A parametric analysis is conducted to determine the determine the optimum coupling gap after the microfluidic channels are loaded with ethanol, as shown optimum coupling gap after the microfluidic channels are loaded with ethanol, as shown in Figure in Figure 6. The return-loss values are enhanced for the ethanol-filled channels. Before the loading 6. The return-loss values are enhanced for the ethanol-filled channels. Before the loading of the of the channels, the electromagnetic (EM) waves pass through air (ε = 1), and the same EM waves channels, the electromagnetic (EM) waves pass through air (εr = 1),r and the same EM waves pass pass through the ethanol-filled PDMS-based channel; thus, the impedance matching is improved (ε of through the ethanol-filled PDMS-based channel; thus, the impedance matching is improved (εrr of ethanol > ε of PDMS > ε of air). G = 0.1 mm is found to be the best impedance-matched case and is r r ethanol > εr of PDMS > εr of air). G = 0.1 mm is found to be the best impedance-matched case and is used for the final design. used for the final design.

Figure investigating the the critical critical coupling coupling gap gap for for the the final finaldesign. design. G G == 0.1 0.1 mm mm Figure 6. 6. Parametric Parametric analysis analysis for for investigating is is found found to to be be the the best best impedance-matched impedance-matched case case under under the the criterion criterion that that the the microfluidic microfluidic channels channels are are loaded with ethanol. loaded with ethanol.

The final layout of the proposed TM02-mode QMSIW resonator with microfluidic channels is The final layout of the proposed TM02 -mode QMSIW resonator with microfluidic channels is shown in Figure 7. shown in Figure 7.

Sensors 2018, 18, 1964

9 of 18

Sensors 2018, 18, x

9 of 18

(a)

(b) Figure 7.7. Layout of the the proposed proposed TM TM 02-mode triangular QMSIW resonator as a dual-detection Figure Layout of 02 -mode triangular QMSIW resonator as a dual-detection chemical sensor: (a) top view; (b) cross-sectional side view. 1 mm-PDMS sheetsheet is considered to model chemical sensor: (a) top view; (b) cross-sectional side view. 1 mm-PDMS is considered to microfluidic channels (h c = 0.6 mm) and an adhesive film (hf = 0.05 mm) is used for bonding purpose. model microfluidic channels (hc = 0.6 mm) and an adhesive film (hf = 0.05 mm) is used for bonding The effective the twoofchannels estimated and foundand to differ as shown purpose. The volumes effective of volumes the two are channels are estimated foundconsiderably, to differ considerably, in shown (a). as in (a).

According to the channel dimensions (see Figure 7), the fluid-carrying capacities (total volume) According to the channel dimensions (see Figure 7), the fluid-carrying capacities (total volume) are estimated as 11.21 and 14.34 µ L for Ch 1 and Ch 2, respectively. The parameters for our final are estimated as 11.21 and 14.34 µL for Ch 1 and Ch 2, respectively. The parameters for our final design design are listed in Table 1. are listed in Table 1. Table1.1.Design Designparameters parametersfor forthe theTM TM20-mode -mode QMSIW QMSIW resonator resonator (unit: (unit: mm). mm). Table 20

Parameter Value Parameter Value Value Parameter Value Ls 30 G 0.1 Ls Ws 30 30 G hs 0.1 1.27 Ws 30 hs 1.27 L 15 a 0.8 L 15 a 0.8 W W 17 17 bb 77 0.7 D D 0.6 0.6 c c 0.7 P P 1.1 1.1 dd 1.1 1.1 Lm Lm 10.510.5 e e 13 13 Wm 1.1 g 8 Wm 1.1 g 8

Parameter

Parameter Parameter i ji j k k ll m m qq h hff hc hc

Value Value 2.8 12.8 1 2.5 2.5 5.5 5.5 0.5 0.5 0.9 0.9 0.05 0.05 0.6 0.6

Note that parameters are already defined according to Figure 7 and the Section 2.

Note that parameters are already defined according to Figure 7 and the section 2.

3. Simulation Analysis 3. Simulation Analysis To analyze the dual-detection capability of the proposed TM02-mode QMSIW resonator as a To analyze the dual-detection capability of the proposed TM02 -mode QMSIW resonator as a microwave sensor, microfluidic channels are characterized using the dielectric properties of empty microwave sensor, microfluidic channels are characterized using the dielectric properties of empty channels and four combinations of ethanol (E) and/or DI water (DI). The simulated return-loss results channels and four combinations of ethanol (E) and/or DI water (DI). The simulated return-loss results are shown in Figure 8. The dielectric properties of the 1-mm PDMS layer were assumed to be εr = 2.7 and tan Δ = 0.05 [41]. The structure resonates at 5.79 GHz when both channels are empty. Full-wave simulations are conducted using a high-frequency structure simulator, and four different resonance

Sensors 2018, 18, 1964

10 of 18

are shown in Figure 8. The dielectric properties of the 1-mm PDMS layer were assumed to be εr = 2.7 and tan ∆ = 0.05 [41]. The structure resonates at 5.79 GHz when both channels are empty. Full-wave Sensors 2018, 18, x 10 of 18 simulations are conducted using a high-frequency structure simulator, and four different resonance frequencies ofof5.44, and 5.32 5.32GHz GHzare are obtained, corresponding toDI], [E, [DI, DI], E], [DI, frequencies 5.44,5.61, 5.61, 5.75, 5.75, and obtained, corresponding to [E, [E,E], E],[E, andE], and [DI, DI], respectively. The dielectric properties of ethanol and DI water are set as ε = 5.08 [DI, DI], respectively. The dielectric properties of ethanol and DI water are set as εr = 5.08r and tanand Δ tan= ∆ = 0.4 [42] and ε = 73 and tan ∆ = 0.3 [43], respectively. After the microfluidic channels were r 0.4 [42] and εr = 73 and tan Δ = 0.3 [43], respectively. After the microfluidic channels were realized, realized, the unloaded Q (simulated) was calculated Q ≈ a43well-known using a well-known the unloaded quality quality factor Qfactor (simulated) was calculated as Q ≈ 43asusing formula formula [44,45].[44,45].

Figure Return loss(simulated) (simulated)for forempty emptychannels channels and and four four possible Figure 8. 8. Return loss possible combinations combinationsofofethanol ethanol(E) (E) and DI water (DI) in the channels, loaded onto the TM 02-mode QMSIW resonator. and DI water (DI) in the channels, loaded onto the TM02 -mode QMSIW resonator.

Sensitivity Analysis Sensitivity Analysis In order to investigate the sensitivity of both channels, a parametric analysis is conducted to In order to investigate the sensitivity of both channels, a parametric analysis is conducted to observe variation in the dielectric properties of the channel material. First, the permittivity variation observe the dielectricinvestigated properties of material. First, variation in eachvariation channel in is individually forthe εr channel values ranging from 2 tothe 10 permittivity with a step size of 2 in (see eachFigure channel is individually investigated for ε values ranging from 2 to 10 with a step sizeand of 2 r 9). Loss tangent (tan δ) is considered as a fixed parameter in both sets of simulations (see Figure 9). Loss tangent (tan δ) is considered as a fixed parameter in both sets of simulations and is arbitrarily chosen as 0.4, which is a common value of the loss tangent of ethanol around 3–5 GHz. is arbitrarily chosen as 0.4,ofwhich is a common value around The dielectric properties the empty channel are set of as the εr = loss 1 andtangent tan δ = of 0 inethanol simulations. Δf3–5 andGHz. Δε The dielectric properties of the empty channel are set as ε = 1 and tan δ = 0 in simulations. ∆f and r are calculated from each pair of resonant frequency and corresponding permittivity values, provided∆ε areincalculated each of resonant frequency corresponding permittivity values, provided the inset from tables in pair Figure 9. For instance, S1 and represents the average sensitivity of Ch 1 uponin thepermittivity inset tables variation in Figureand 9. For instance, S1 is estimated as represents follows: the average sensitivity of Ch 1 upon permittivity variation and is estimated as follows:

S1= ∆f

S1 =

Δf Δf1 +Δf 2 +Δf3 +Δf 4 10+20+0+10 =∆f +∆f +∆f +∆f =10 + 20 + 0 + 10= 6.25MHz/ε r 2 +Δε 3 1 +Δε 4 Δε 2+2+2+2 = Δε = 6.25MHz/εr 1 2 3 +Δε= 4

∆ε ∆ε1 +∆ε2 +∆ε3 +∆ε4 2+2+2+2 With the data taken from Figure 9b and applying the above equation, the average sensitivity of the permittivity data taken from Figure and applying equation, the average sensitivity of ChWith 2 upon variation is 9b calculated as S2 =the 2.5above MHz/ε r. A higher sensitivity of Ch 1 is Chperhaps 2 upondue permittivity variation is calculated as S2 = 2.5 MHz/ε . A higher sensitivity of Ch 1 is to more overlapped area containing a stronger E-field.r perhapsNow, due to overlapped areatangent containing a stronger E-field. themore influence of the loss inside each channel (material) on the sensitivity of our Now, the influence of the loss tangent inside each channel of our proposed sensor is investigated (see Figure 10). Loss tangent (tan (material) δ from 0 to on 0.5 the withsensitivity an incremental proposed sensor is investigated (see Figure 10). Loss tangent (tan δ from 0 to 0.5 with an incremental step of 0.1) and a fixed permittivity value (arbitrarily chosen as εr = 10) are assigned to Ch 1 while the step and a fixed value (arbitrarily chosenthe as loss εr = tangent 10) are assigned to Ch 1 while Chof2 0.1) is considered as permittivity empty, and vice versa. As we know, is associated with the thereturn Ch 2 is considered as for empty, and vice versa.loss As of weCh know, the from loss tangent associated with loss magnitude; instance, the return 1 varies 8.5 dB tois12.8 dB when tan the δ changes 0 to 0.5, for while the resonant frequency almost constant with slight of 10 return lossfrom magnitude; instance, the return loss of Ch 1remains varies from 8.5 dB toa12.8 dBshift when tan MHz (see inset0table provided in Figure 10a). From Figure 10b,remains the return loss changes 11shift dB toof δ changes from to 0.5, while the resonant frequency almost constant with afrom slight 47 dB when tan δ varies from 0 to 0.5. An equal amount of incremental change in tan δ causes a larger variation in the return loss (magnitude) of Ch 2 as compared with Ch 1, which can be explained by reasoning that the volume of Ch 2 is higher than that of Ch 1.

Sensors 2018, 18, 1964

11 of 18

10 MHz (see inset table provided in Figure 10a). From Figure 10b, the return loss changes from 11 dB to 47 dB when tan δ varies from 0 to 0.5. An equal amount of incremental change in tan δ causes a larger variation in the return loss (magnitude) of Ch 2 as compared with Ch 1, which can be explained by reasoning that the volume of Ch 2 is higher than that of Ch 1. Sensors 2018, 18, x 11 of 18

(a)

(b)

Figure 9. 9. Effect Effectofofpermittivity permittivityvariation variationofofmaterial materialinside insidethe the channel channelon on the the sensitivity sensitivity of of the the Figure proposedsensor. sensor.(a)(a) is empty, permittivity variation 1 is analyzed. Ch 1 is proposed ChCh 2 is2 empty, andand permittivity variation in Chin1 Ch is analyzed. (b) Ch 1(b) is empty, empty, and permittivity inanalyzed. Ch 2 is analyzed. Loss value tangent value is considered as a constant and permittivity variationvariation in Ch 2 is Loss tangent is considered as a constant (tan δ = (tan δ = 0.4) during these sets of simulations. 0.4) during these sets of simulations.

(a)

(b)

Figure Figure 10. 10. Effect Effectofofloss losstangent tangentvariation variationof ofmaterial materialinside insidethe the channel channel on on the the sensitivity sensitivity of of the the proposed proposedsensor. sensor.(a) (a)Ch Ch22isisempty, empty,and andloss losstangent tangentvariation variationin inCh Ch11isisanalyzed. analyzed.(b) (b)Ch Ch11isisempty, empty, andloss losstangent tangentvariation variation in in Ch Ch 22 is is analyzed. analyzed. Permittivity Permittivity is is considered considered as as a constant constant (εrr ==10) 10)during during and thesesets setsof ofsimulations. simulations. these

4.4.Fabrication Fabricationand andMeasurement Measurement 4.1.Fabrication Fabrication 4.1. Theconductive conductivepattern patternand andground ground are are realized realized on on the the top top and bottom of a Rogers RT/duroid The RT/duroid 6010LM substrate using conventional photolithography. Unlike our previous dual-detection 6010LM substrate using Unlike our previous dual-detection microwave sensor, this study is based on a single substrate integrated with two asymmetric microfluidic channels loaded in two E-field regions. The two PDMS-based microfluidic channels were fabricated using a laser cutting machine, which was a quick process, although the channel surfaces are slightly rough. The fabricated prototype resonator, the channels engraved inside the PDMS slabs, and the demonstration of the fluid-carrying capability are presented in Figure 11. An

Sensors 2018, 18, 1964

12 of 18

microwave sensor, this study is based on a single substrate integrated with two asymmetric microfluidic channels loaded in two E-field regions. The two PDMS-based microfluidic channels were fabricated using a laser cutting machine, which was a quick process, although the channel surfaces are slightly rough. The fabricated prototype resonator, the channels engraved inside the PDMS slabs, and the demonstration of the fluid-carrying capability are presented in Figure 11. An adhesive film is used to bond the PDMS-based channels onto the conductive pattern, which is realized on duroid 6010LM, as shown in Figure 11e.

Sensors 2018, 18, x

12 of 18

(a)

(b)

(c)

(d)

(e) Figure 11. (a) Fabricated Fabricated prototype prototype integrated views of Figure 11. (a) integrated with with microfluidic microfluidic channels. channels. (a,b) (a,b) Top/bottom Top/bottom views of the fabricated prototype. (c) Both the channels realized using a laser cutting machine. (d) Testing the fabricated prototype. (c) Both the channels realized using a laser cutting machine. (d) Testing of of the channels channels by by injecting injecting colored colored water water for for verification (only Ch is shown). shown). (e) (e) Fabricated Fabricated prototype prototype the verification (only Ch 22 is resonator integrated resonator integrated with with microfluidic microfluidic channels. channels.

4.2. Measurement The fabricated prototype integrated with microfluidic channels is connected to a vector network analyzer (Anritsu MS2038C, MS2038C, manufactured manufactured by by Anritsu Anritsu Corporation Corporation Kanagawa Kanagawa Prefecture, Prefecture, Japan). Japan). analyzer The measurement setup is shown in Figure 12a. The return loss is measured when the channels are empty and compared with the simulation results, as shown in Figure 12b. The simulated and measured resonance frequencies (5.81 GHz) exhibit excellent agreement. To demonstrate the dual-detection capability, ethanol and DI water are alternately injected into the channels, in four possible combinations (see Figure 13). The resonance frequency and return loss in each case obtained from the simulation and measurement are compared in Table 2. To evaluate the accuracy of the measurement, the relative error in the resonance frequency is calculated (referring

Sensors 2018, 18, 1964

13 of 18

empty and compared with the simulation results, as shown in Figure 12b. The simulated and measured resonance frequencies (5.81 GHz) exhibit excellent agreement. To demonstrate the dual-detection capability, ethanol and DI water are alternately injected into the channels, in four possible combinations (see Figure 13). The resonance frequency and return loss in each case obtained from the simulation and measurement are compared in Table 2. To evaluate the accuracy of the measurement, the relative error in the resonance frequency is calculated (referring to Equation (3) from [34]). Sensors 2018, 18, x 13 of 18

(a)

(b) Figure 12. 12.(a) (a) Measurement setup: the proposed TM02triangular -mode triangular QMSIW resonator Figure Measurement setup: the proposed TM02-mode QMSIW resonator (without (without loading of microfluidic channels) is measured using a vector network analyzer and resonates loading of microfluidic channels) is measured using a vector network analyzer and resonates at 5.81 at 5.81 (b) Simulated and measured valuesfor obtained for the proposed resonator GHz. (b)GHz. Simulated and measured return-lossreturn-loss values obtained the proposed resonator without the without of themicrofluidic loading of microfluidic channels. The unloaded simulatedQunloaded Q ≈ 43 is calculated loading channels. The simulated ≈ 43 is calculated according according to a 3-dB to a 3-dB bandwidth at the resonance bandwidth at the resonance frequency.frequency.

Table 2. Comparison between the simulation and measurement results after the loading of microfluidic channels. The resonant frequencies and return-loss measurements are provided, corresponding to empty channels and the four possible combinations of ethanol and DI water. Ch 1, Ch 2 Air, Air Ethanol, DI water DI water, Ethanol Ethanol, Ethanol DI water, DI water

Simu. fr (GHz) 5.79 5.44 5.61 5.75 5.32

Simu. S11 (dB) −7.10 −6.14 −18.81 −19.89 −5.99

Meas. fr (GHz) 5.791 5.445 5.614 5.760 5.321

Meas. S11 (dB) −7.14 −6.74 −17.28 −25.71 −6.74

Relative Error in fr (%) 0.02 0.09 0.07 0.17 0.02

Sensors 2018, 18, 1964

14 of 18

Sensors 2018, 18, x

14 of 18

Figure 13. 13. Comparison Comparisonof ofthe thereturn-loss return-lossvalues valuesobtained obtained from simulation and measurement Figure from thethe simulation and measurement for for all all possible combinations of ethanol DI water for empty channels. possible combinations of ethanol andand DI water andand for empty channels.

5. Discussion Table 2. Comparison between the simulation and measurement results after the loading of microfluidic channels. Thethe resonant frequencies and return-loss are (frequency provided, corresponding to To evaluate proposed microwave resonator,measurements its performance shifts and electrical empty channels and the four possible combinations of ethanol and DI water. size) is compared with that of recently published SIW resonators, as shown in Table 3. Our

microwave resonator exhibits comparable performance; moreover, the size is reduced. Notably, the Ch 1, Ch 2 Simu. fr (GHz) Simu. S11 (dB) Meas. fr (GHz) Meas. S11 (dB) Relative Error in fr (%) applications of the other SIW resonators in Table 3 are limited to a single analyte. Air, Air 5.79 −7.10 5.791 −7.14 0.02 Ethanol, DI water 5.44 −6.14 5.445 −6.74 0.09 3. Comparison5.61 of our TM 02-mode with recently chemical DI Table water, Ethanol −18.81QMSIW resonator 5.614 −17.28 proposed SIW0.07 Ethanol, Ethanol 5.75 −19.89 5.760 −25.71 0.17 sensors. DI water, DI water 5.32 −5.99 5.321 −6.74 0.02

Ref. fo (GHz) Δfmax * (MHz) Technology Electrical Length† λg × λg Sensing This work 5.81 470 QMSIW 1.62 × 1.07 Dual 5. Discussion [14] 4.65 400 EMSIW 0.94 × 0.9 Single To evaluate microwave resonator, shiftsSingle and electrical [40] the proposed 17.08 610 SIWits performance 3 ×(frequency 2.61 [44] with that 5 of recently 380published SIW SIWresonators, as1.85 × 1.85in Table 3. Our Single size) is compared shown microwave [46] comparable 13.48 170 SIW the size is reduced. 2.26 × 1.92 resonator exhibits performance; moreover, Notably, the Single applications of *ΔfmaxSIW represents the maximum shiftto(absolute of the device among all the test cases. the other resonators in Table 3frequency are limited a singlevalue) analyte.

The performance of our waswith compared that ofSIW existing Table 3. Comparison of dual-detection our TM02 -modechemical QMSIW sensor resonator recentlywith proposed microwave detection sensors, as shown in Table 4. To evaluate the performance of chemicaldual/multiple sensors.

f Sensing 100 , where fo

these sensors, frequency shift was calculated, which is defined Ref. thefofractional (GHz) ∆fmax * (MHz) Technology Electrical Length†asλg S×λg This work

5.81

470

QMSIW

1.62 × 1.07

[46]

13.48

170

SIW

2.26 × 1.92

Dual

Δf represents frequency is the operating frequency. The unit of theSingle sensitivity [14] the absolute 4.65 400 shift and fo EMSIW 0.94 × 0.9 S is (%).[40] A qualitative our proposed sensor existing multichannel 17.08analysis of 610 SIW compared with 3× 2.61 Singlesensors [44] in Table55. 380 SIW 1.85 × 1.85 Single is provided Single

* ∆f4.max represents thecomparison maximum frequency value) the device among resonator all the test cases. Table Performance betweenshift our(absolute proposed TMof02-mode QMSIW and other microwave dual/multichannel sensors.

The performance of our dual-detection chemical sensor was compared with that of existing Ref. fo (GHz) Δf * (MHz) S (%) Physical Size (mm ×mm) Electrical Size† (λg × λg) microwave dual/multiple detection sensors, as shown in Table 4. To evaluate the performance of these [34] 8 430 5.37 60 × 40 2.37 ∆ f × 1.58 sensors, the fractional frequency shift was calculated, which is defined as S = f o × 100, where ∆f [30] 3 170 5.66 35 × 32 1.12 × 1.02 represents the absolute frequency shift and6.15 fo is the operating frequency. The unit 1.13 of the sensitivity S is [31] 6.5 400 30 × 22 × 0.86 [32] [33] This work

0.87 2 5.77

110 200 31

12.6 10 8.1

86 × 62 50 × 40 30 × 30

0.8 × 0.57 1.4 × 1.12 1.62 × 1.07

Sensors 2018, 18, 1964

15 of 18

(%). A qualitative analysis of our proposed sensor compared with existing multichannel sensors is provided in Table 5. Table 4. Performance comparison between our proposed TM02 -mode QMSIW resonator and other microwave dual/multichannel sensors. Ref.

fo (GHz)

∆f * (MHz)

S (%)

Physical Size (mm × mm)

Electrical Size† (λg × λg )

[34] [30] [31] [32] [33] This work

8 3 6.5 0.87 2

430 170 400 110 200

5.37 5.66 6.15 12.6 10

60 × 40 35 × 32 30 × 22 86 × 62 50 × 40

2.37 × 1.58 1.12 × 1.02 1.13 × 0.86 0.8 × 0.57 1.4 × 1.12

5.77

31

8.1

30 × 30

1.62 × 1.07

* ∆f represents the absolute frequency shift when frequencies corresponding to ethanol/DI water are compared with that for the empty channel. Conventionally, frequency shifts in RF chemical sensors are compared with reference to DI water. However, because of the insufficient data in some reports, air was used as the reference media. Some studies, such as [30], demonstrated multichannel sensing without the utilization of microfluidic channels.

Table 5. Qualitative features of our proposed TM02 -mode QMSIW resonator along with other microwave chemical sensors capable of dual/multiple detection. Ref.

Technology

Configuration

This work [34] [30] [31] [32] [33]

TM02 -mode QMSIW TE20 -mode SIW MM MM MM Dual-mode resonator

Unit cell on single substrate Stacked layer Array Array Array Stacked layer

Noncontact Independent Tuning Yes Yes No Yes Yes Yes

No No Yes Yes No Yes

Detection Dual Dual Multiple Dual Partially Dual Dual

Note: Topologies in [30–32] are 4 SRRs, 3 open split ring resonators (OSRRs), and 2 SRRs, respectively, whereas MM represents metamaterial technology.

Misalignment of the channels in the fabricated prototype may lead to inaccurate measurements and a compromised sensitivity [47]. The authors consider that such misalignment cannot be eradicated via manual handling; however, it can be avoided to some extent using alignment marks on the sensor chip. To investigate this, the dimensions of PDMS slabs (containing channels) were designed to cause overlap along the width of substrate or along one side of the QMSIW patch (as discussed in detail in Section 2). As indicated by Table 3, single-analyte sensors may suffer in certain scenarios, as discussed in our previous work [34]. A dual/multiple sensor similar to that proposed herein can provide reliable detection in the case where one channel is biased [48,49]. In [30], a microwave resonator capable of detecting four different chemicals is presented; however, it is a direct-contact sensor, which may raise concerns such as contamination, as discussed in [34]. These issues are prevented in our proposed microfluidic integrated microwave resonator. Among the dual/multiple detection microwave sensors compared in Table 4, only those from [30,33] exhibit independent tuning capability. The tuning of the resonance frequencies in our proposed sensor is partially independently-tunable, owing to its inherent nature (two E-field regions tend to control a single resonance). The sensor proposed in [33] is comprised of five stacked layers constructed using low-temperature cofired ceramic technology and laser micromachining. RF/microwave chemical sensors suffer from being nonselective. To resolve this issue, hybrid sensors (RF sensor with additional coatings) have been proposed [49,50]. Every sensor is designed to meet certain performance criteria or optimized to target a specific application. Our proposed sensor has a simple design and is fabricated using conventional lithography; thus, it is inexpensive. Being noncontact and having sensitivity comparable to that of

Sensors 2018, 18, 1964

16 of 18

contemporary microwave sensors, the TM02 -mode QMSIW resonator has potential as a dual-detection microwave sensor. 6. Conclusions A TM02 -mode triangular QMSIW resonator is proposed for the detection of two chemicals regardless of their dielectric constants. The novelty of the proposed resonator is its dual-detection capability, and size reduction is achieved using the QMSIW approach. Loading two asymmetric channels in two distinct E-field regions was insufficient to obtain different resonance frequencies. A certain difference in the effective volume of each channel must exist in order to obtain the unique resonance regardless of the dielectric constants of the channel loadings. To demonstrate the potential of the proposed dual-detection chemical sensor, ethanol and DI water were injected into the microfluidic channels, and return-loss measurements agreed well with simulations. The performance of the proposed sensor can be improved with regard to the sensitivity and size reduction. To improve the sensitivity of the dual-detection sensor, the overlapped area of both channels can be increased, for example, by using closely spaced meander-shaped channels. Further miniaturization can be achieved, for example, by using a sixteen-mode SIW resonator. Author Contributions: S.L. conceived the idea; A.S. performed the design, fabrication, and measurement and wrote the manuscript. S.L. revised the manuscript. Funding: This work was supported by the National Research Foundation of Korea, with a grant funded by the Korean government (No. 2017R1A2B3003856). Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5.

6.

7. 8.

9. 10. 11. 12.

Stich, M.I.J.; Fischer, L.H.; Wolfbeis, O.S. Multiple fluorescent chemical sensing and imaging. Chem. Soc. Rev. 2010, 39, 3102–3114. [CrossRef] [PubMed] Su, J.; Goldberg, A.F.; Stoltz, B.M. Label-free detection of single nanoparticles and biological molecules using microtoroid optical resonators. Light Sci. Appl. 2016, 5, 1–6. [CrossRef] Cheng, D.K. Field and Wave Electromagnetics, 2nd ed.; Addison Wesley Inc.: Boston, MA, USA, 1989; ISBN 0-201-01239-1. Kang, H.; Lim, S. Enhanced-gain planar substrate-integrated waveguide cavity-backed slot antenna with rectangular slot window on superstrate. ETRI J. 2014, 36, 1062–1065. [CrossRef] Park, W.; Lim, S. A Low Phase-Noise Microwave Oscillator Using a Substrate Integrated Waveguide Resonator based on Complementary Split Ring Resonator. In Proceedings of the Asia-Pacific Microwave Conference 2011, Melbourne, VIC, Australia, 5–8 December 2011; pp. 371–374. Eom, D.; Lee, H. A Broadband Half-Mode Substrate Integrated Waveguide Quadrature Wilkinson Power Divider Using Composite Right/Left-Handed Transmission Line. J. Electromagn. Eng. Sci. 2017, 17, 9–13. [CrossRef] El Matbouly, H.; Boubekeur, N.; Domingue, F. Passive Microwave Substrate Integrated Cavity Resonator for Humidity Sensing. IEEE Trans. Microw. Theory Tech. 2015, 63, 4150–4156. [CrossRef] Ndoye, M.; El Matbouly, H.; Sama, Y.N.; Deslandes, D.; Domingue, F. Sensitivity evaluation of dielectric perturbed substrate integrated resonators for hydrogen detection. Sens. Actuators A Phys. 2016, 251, 198–206. [CrossRef] Bozzi, M.; Georgiadis, A.; Wu, K. Review of substrate-integrated waveguide circuits and antennas. IET Microw. Antennas Propag. 2011, 5, 909–920. [CrossRef] Liu, B.; Hong, W.; Wang, Y.Q.; Lai, Q.H.; Wu, K. Half Mode Substrate Integrated Waveguide (HMSIW) 3-dB coupler. IEEE Microw. Wirel. Compon. Lett. 2007, 17, 22–24. [CrossRef] Dong, Y.; Itoh, T. Composite Right/Left-Handed Substrate Integrated Waveguide and Half Mode Substrate Integrated Waveguide Leaky-Wave Structures. IEEE Trans. Antennas Propag. 2011, 59, 767–775. [CrossRef] Memon, M.U.; Lim, S. Frequency-tunable compact antenna using quarter-mode substrate integrated waveguide. IEEE Antennas Wirel. Propag. Lett. 2015, 14, 1606–1609. [CrossRef]

Sensors 2018, 18, 1964

13. 14. 15.

16. 17. 18. 19. 20.

21. 22.

23. 24. 25. 26. 27. 28. 29. 30. 31.

32.

33. 34. 35. 36.

17 of 18

Eom, S.; Memon, M.; Lim, S. Frequency-Switchable Microfluidic CSRR-Loaded QMSIW Band-Pass Filter Using a Liquid Metal Alloy. Sensors 2017, 17, 699. [CrossRef] [PubMed] Seo, Y.; Memon, M.U.; Lim, S. Microfluidic Eighth-Mode Substrate-Integrated- Waveguide Antenna for Compact Ethanol Chemical Sensor Application. IEEE Trans. Antennas Propag. 2016, 64, 3218–3222. [CrossRef] Sam, S.; Lim, S. Electrically small eighth-mode substrate-integrated waveguide (EMSIW) antenna with different resonant frequencies depending on rotation of complementary split ring resonator. IEEE Trans. Antennas Propag. 2013, 61, 4933–4939. [CrossRef] Azad, A.R.; Mohan, A. A compact sixteenth-mode substrate integrated waveguide bandpass filter with improved out-of-band performance. Microw. Opt. Technol. Lett. 2017, 59, 1728–1733. [CrossRef] Jones, T.R.; Zarifi, M.H.; Daneshmand, M. Miniaturized Quarter-Mode Substrate Integrated Cavity Resonators for Humidity Sensing. IEEE Microw. Wirel. Compon. Lett. 2017, 27, 612–614. [CrossRef] Jin, C.; Li, R.; Alphones, A.; Bao, X. Quarter-mode substrate integrated waveguide and its application to antennas design. IEEE Trans. Antennas Propag. 2013, 61, 2921–2928. [CrossRef] Jin, C.; Shen, Z.; Member, S. Compact Triple-Mode Filter Based on Quarter-Mode Substrate Integrated Waveguide. IEEE Trans. Microw. Theory Tech. 2014, 62, 37–45. [CrossRef] Zarifi, M.H.; Sadabadi, H.; Hejazi, S.H.; Daneshmand, M.; Sanati-Nezhad, A. Noncontact and Nonintrusive Microwave-Microfluidic Flow Sensor for Energy and Biomedical Engineering. Sci. Rep. 2018, 8, 139. [CrossRef] [PubMed] Ghosh, S.; Srivastava, K.V. An Angularly Stable Dual-Band FSS with Closely Spaced Resonances Using Miniaturized Unit Cell. IEEE Microw. Wirel. Compon. Lett. 2017, 27, 218–220. [CrossRef] Trungo, N.T.; Lim, S. Metamaterial Absorber using Complementary Circular Sector Resonator. In Proceedings of the 2016 International Symposium on Antennas and Propagation (ISAP), Okinawa, Japan, 24–28 October 2016; pp. 176–177. Smith, D.R.; Smith, D.R.; Pendry, J.B.; Wiltshire, M.C.K. Metamaterials and negative refractive index. Science 2004, 305, 788–792. [CrossRef] [PubMed] Choi, S.; Salim, A.; Jeong, H.; Lim, S. High-efficiency and compact metamaterial-inspired 900 MHz rectifier. J. Microw. Power Electromagn. Energy 2016, 50, 168–181. [CrossRef] Kim, G.; Lee, B. Synthesis of Bulk Medium with Negative Permeability Using Ring Resonators. J. Electromagn. Eng. Sci. 2016, 16, 67–73. [CrossRef] Sadeqi, A.; Sonkusale, S. Low-cost metamaterial-on-paper chemical sensor. Optics Express 2017, 25, 16092–16100. [CrossRef] [PubMed] Lim, S.; Caloz, C.; Itoh, T. Metamaterial-Based Electronically Controlled Transmission-Line Structure as a Novel Leaky-Wave Antenna With Tunable. IEEE Trans. Microw. Theory Tech. 2004, 52, 2678–2690. [CrossRef] Eom, S.; Lim, S. Stretchable complementary split ring resonator (CSRR)-based radio frequency (RF) sensor for strain direction and level detection. Sensors 2016, 16, 1667. [CrossRef] [PubMed] Choi, S.; Eom, S.; Tentzeris, M.M.; Lim, S. Inkjet-Printed Electromagnet-Based Touchpad Using Spiral Resonators. J. Microelectromech. Syst. 2016, 25, 947–953. [CrossRef] Withayachumnankul, W.; Jaruwongrungsee, K.; Fumeaux, C.; Abbott, D. Metamaterial-Inspired Multichannel Thin-Film Sensor. IEEE Sens. J. 2012, 12, 1455–1458. [CrossRef] Byford, J.A.; Park, K.Y.; Chahal, P. Metamaterial inspired periodic structure used for microfluidic sensing. In Proceedings of the IEEE 65th Electronic Components and Technology Conference, San Diego, CA, USA, 26–29 May 2015; pp. 1997–2002. Velez, P.; Su, L.; Grenier, K.; Mata-Contreras, J.; Dubuc, D.; Martin, F. Microwave Microfluidic Sensor Based on a Microstrip Splitter/Combiner Configuration and Split Ring Resonators (SRRs) for Dielectric Characterization of Liquids. IEEE Sens. J. 2017, 17, 6589–6598. [CrossRef] Jankovic, N.; Radonic, V. A microwave microfluidic sensor based on a dual-mode resonator for dual-sensing applications. Sensors 2017, 17, 2713. [CrossRef] [PubMed] Salim, A.; Memon, M.; Lim, S. Simultaneous Detection of Two Chemicals Using a TE20-Mode Substrate-Integrated Waveguide Resonator. Sensors 2018, 18, 811. [CrossRef] [PubMed] Salim, A.; Lim, S. Review of Recent Metamaterial Microfluidic Sensors. Sensors 2018, 18, 232. [CrossRef] [PubMed] Salim, A.; Lim, S. Review of recent inkjet-printed capacitive tactile sensors. Sensors 2017, 17, 2593. [CrossRef] [PubMed]

Sensors 2018, 18, 1964

37.

38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

50.

18 of 18

Yun, T.; Lim, S. High-Q and miniaturized complementary split ring resonator-loaded substrate integrated waveguide microwave sensor for crack detection in metallic materials. Sens. Actuators A Phys. 2014, 214, 25–30. [CrossRef] Guo, Z.; Chin, K.S.; Che, W.; Chang, C.C. Cross-coupled bandpass filters using QMSIW cavities and S-shaped slot coupling structures. J. Electromagn. Waves Appl. 2013, 27, 160–167. [CrossRef] Memon, M.U.; Lim, S. Review of reconfigurable substrate-integrated-waveguide antennas. J. Electromagn. Waves Appl. 2014, 28, 1815–1833. [CrossRef] Memon, M.U.; Lim, S. Millimeter-wave chemical sensor using substrate-integrated-waveguide cavity. Sensors 2016, 16, 1829. [CrossRef] [PubMed] Salim, A.; Lim, S. Complementary Split-Ring Resonator-Loaded Microfluidic Ethanol Chemical Sensor. Sensors 2016, 16, 1802. [CrossRef] [PubMed] Meissner, T.; Wentz, F.J. The complex dielectric constant of pure and sea water from microwave satellite observations. IEEE Trans. Geosci. Remote Sens. 2004, 42, 1836–1849. [CrossRef] Gregory, A.P.; Clarke, R.N. Tables of the Complex. Permittivity of Dielectric Reference Liquids at Frequencies up to 5 GHz; NPL Report MAT 23; National Physical Laboratory: Teddington, UK, 2012. Memon, M.U. Microfluidic High-Q Circular Substrate-Integrated Waveguide (SIW) Cavity for Radio Frequency (RF) Chemical Liquid Sensing. Sensors 2018, 18, 143. [CrossRef] [PubMed] Pozar, D.M. Microwave Engineering; Wiley: Hoboken, NJ, USA, 1998; ISBN 0-471-17096-8. Salim, A.; Kim, S.; Park, J.Y.; Lim, S. Microfluidic Biosensor Based on Microwave Substrate Integrated Waveguide Cavity Resonator. J. Sens. 2018, 2018, 1324145. [CrossRef] Awang, R.A.; Tovar-Lopez, F.J.; Baum, T.; Sriram, S.; Rowe, W.S.T. Meta-atom microfluidic sensor for measurement of dielectric properties of liquids. J. Appl. Phys. 2017, 121, 094506. [CrossRef] Swiontek, S.E.; Pulsifer, D.P.; Lakhtakia, A. Optical sensing of analytes in aqueous solutions with a multiple surface-plasmon-polariton-wave platform. Sci. Rep. 2013, 3, 1409. [CrossRef] [PubMed] Zarifi, M.H.; Farsinezhad, S.; Wiltshire, B.D.; Abdorrazaghi, M.; Mahdi, N.; Kar, P.; Daneshmand, M.; Shankar, K. Effect of phosphonate monolayer adsorbate on the microwave photoresponse of TiO2 nanotube membranes mounted on a planar double ring resonator. Nanotechnology 2016, 27, 375201. [CrossRef] [PubMed] Ali, M.A.; Cheng, M.M.C.; Chen, J.C.M.; Wu, C.-T.M. Microwave Gas Sensor based on Graphene-loaded Substrate Integrated Waveguide Cavity Resonator. In Proceedings of the IEEE MTT-S International Microwave Symposium (IMS), San Francisco, CA, USA, 22–27 May 2016; pp. 4–7. © 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).