Planar Microstrip Ring Resonators for Microwave-Based Gas ... - MDPI

sensors Article

Planar Microstrip Ring Resonators for Microwave-Based Gas Sensing: Design Aspects and Initial Transducers for Humidity and Ammonia Sensing Andreas Bogner, Carsten Steiner, Stefanie Walter, Jaroslaw Kita, Gunter Hagen and Ralf Moos * Department of Functional Materials, University of Bayreuth, 95447 Bayreuth, Germany; [email protected] (A.B.); [email protected] (C.S.); [email protected] (S.W.); [email protected] (J.K.); [email protected] (G.H.) * Correspondence: [email protected]; Tel.: +49-921-55-7401 Received: 21 September 2017; Accepted: 17 October 2017; Published: 24 October 2017

Abstract: A planar microstrip ring resonator structure on alumina was developed using the commercial FEM software COMSOL. Design parameters were evaluated, eventually leading to an optimized design of a miniaturized microwave gas sensor. The sensor was covered with a zeolite film. The device was successfully operated at around 8.5 GHz at room temperature as a humidity sensor. In the next step, an additional planar heater will be included on the reverse side of the resonator structure to allow for testing of gas-sensitive materials under sensor conditions. Keywords: microwave cavity perturbation; resonant frequency; radio frequency based gas sensors; zeolites; in operando spectroscopy

1. Introduction Sensors for gas detection include optical gas sensors based on light absorption, metal oxide gas sensors based on a resistive effect (chemiresistors), catalytic gas sensors through adsorption and thermic reaction, gravimetric SAW detectors, gas chromatography, calorimetric devices, biochemical sensors, and many other sensors based on capacitive, amperometric, or potentiometric effects [1]. In typical chemiresistors, gas-sensitive functional films are applied on substrates that are covered with (interdigital) electrodes, and their resistances or their complex impedances are determined [2–4]. Mostly, sensors are operated in the range of room temperature to 400 ◦ C. High ohmic electrode–film interfaces and/or very high resistivities of the sensitive materials may exclude promising sensor materials from technical application. As an alternative approach, devices based on microwave transducers have also been suggested with growing research interest in recent years [5–9]. For sensor purposes, microwaves, especially electromagnetic waves in the range from 1 to 20 GHz, are typically used with planar or hollow waveguides. The sensor effect is the analyte-dependent permittivity of the material. With respect to miniaturization and low-cost applications, planar waveguide-based chemical sensors have attracted recent attention [10]. Several microwave-based sensors have been proposed in the past: moisture detection for the food and chemical industries [11–14] or glucose detection for biomedical applications [15], to name a few. In the past few years, planar microwave transducers for NH3 , toluene, CO2 , or CH4 based on the highly sensitive resonant method were suggested [5–9]. Rossignol et al. used coplanar microwave transducers and developed a resonant coplanar structure for detecting NH3 and toluene in concentrations up to 500 ppm [7]. They carried out their experiments at room temperature and regenerated the sensitive phthalocyanine layer under Sensors 2017, 17, 2422; doi:10.3390/s17102422

www.mdpi.com/journal/sensors

Sensors 2017, 17, 2422

2 of 14

argon. Bailly et al. demonstrated a hematite- and a zeolite-based variant of this sensor [8,9]. A concept by Zarifi et al. involves planar resonators and an active feedback loop for a gain that yields up to five times higher quality factors and thus higher resolutions [16]. They also published a zeolite-based version of their high Q resonator for sensing of CH4 and CO2 concentrations between 1% and 50% [5]. With respect to the sensing aspect, framework materials that adsorb chemical species, like metal-organic frameworks (MOFs) [17] or zeolites [18–20], may be preferred for microwave gas sensors range since the adsorption of large amounts of polar gas species goes along with an appropriate change in the complex electrical permittivity. Very recently, Bahoumina et al. proposed a chemical gas sensor with a planar capacitive microwave transducer in the microwave range and a polymer carbon nanomaterial as the sensitive layer [21], and Zarifi et al. demonstrated their active-resonator concept with MOFs as a sensitive material to detect CO2 [22]. Another, newer publication of Bailly et al. demonstrated a microstrip spiral resonator covered with titanium dioxide nanoparticles [8]. Recently, further microwave-based sensing concepts with different sensing purposes and planar geometries were published. They include a coupled line section for dielectric sample measurements, interdigital capacitors for liquid mixture concentration measurement, and a disc-shaped tip with a zeolite for thermal mass gas sensing [23–26]. Even proteins can be detected if the resonator (or the split in a split ring resonator) is functionalized with aptamers [27]. Another biomedical application using an interdigital microstrip capacitor is reported by Rydosz et al. They developed a microwave-based sensor to detect various volatile organic compounds and confirmed that it can be used for exhaled acetone detection [28]. The previous solutions demonstrate the capability of microwave-based sensors to monitor gas concentrations and/or the analyte loading of a sensitive layer. However, research on design and application of microwave transducers for gas detection is in its infancy and needs further investigations to develop integrated sensor devices. In addition, the size of recent devices prevents small applications and dynamic fast measurements with gas sensitive framework layers like zeolites due to their room temperature application. Hence, in this work, a very small 9 GHz resonant microstrip structure with a ring geometry for NH3 detection is investigated. As the sensitive material, the ring resonator is coated with an NH3 and water adsorbing Fe-zeolite as it is used for selective catalytic NOx reduction reactions in automotive exhausts [29,30]. It is demonstrated that small humidity and ammonia concentrations can be detected. The paper discusses design aspects that were obtained by FEM simulation for planar cavity ring resonators and demonstrates their applicability by fabricating an entire setup and conducting initial measurements at room temperature. We expect that the proposed microwave transducer will open up new possibilities for material characterization and gas-sensing purposes, especially when in the second step a heater layer will be implemented. 2. Principle of Planar Microwave Transducers Due to their interaction with the analyte, the gas sensitive material changes its complex electrical properties of permittivity ε = ε’ − jε” (with j being the imaginary unit) and permeability µ = µ’ − jµ”. The varying material properties lead to a changing wave propagation. Besides these intrinsic material effects, the wave propagation depends on the material’s geometry and on the design of the used waveguide (extrinsic properties) [10]. Generally, electromagnetic waves are guided to a desired transmission mode by restricting their expansion in one or two dimensions. A widely used transmission structure is the planar microstrip line, as depicted in the Appendix A (Figure A1). Microstrip lines consist of a strip conductor and a ground plane separated by a dielectric substrate and thus only support the transversal electromagnetic mode (TEM). TEM modes can be found whenever there are two conductors in only one medium. Hence, to be more precise, the supported wave mode for microstrips is not strictly TEM since the waves propagate not only in the substrates but also in the medium above the conductor line. This causes different phase velocities and, in turn, a longitudinal component of the electric magnetic field. However, since this component is very small it can be often neglected and the supported wave mode is then called

Sensors 2017, 17, 2422

3 of 14

c

s

ε

f f e

h Lc

=

e fr

quasi-TEM mode. In addition, the two-dimensional structure of microstrips make them well suited for miniaturization and integration with other components and, as a result of the single plane structure, they canSensors be fabricated 2017, 17, 2422conventionally by thick or thin film technology, photolithography, photoetching, 3 of 14 or laser structuring [31]. as for a result of the single plane structure, they can be fabricated by thick or thin film Like microwave-based material characterization methodsconventionally based on microstrips, non-resonant technology, photolithography, photoetching, or laser structuring [31]. or resonant methods are applicable, but resonant methods are preferred for sensing applications Like for microwave-based material characterization methods based on microstrips, nondue to their higher sensitivity and accuracy. Well-known in this respect is the resonant perturbation resonant or resonant methods are applicable, but resonant methods are preferred for sensing method,applications which is based ontheir the resonant frequency of the scatteringinparameters introducing due to higher sensitivity andchange accuracy. Well-known this respect by is the resonant a sample perturbation [10]. Examples for such applications in the field of chemical sensors are the characterization method, which is based on the resonant frequency change of the scattering parameters of catalystby powder samples in situ during chemical reactions [32,33], or the monitoring of are catalysts introducing a sample [10]. Examples for such applications in the fieldstate of chemical sensors the directlycharacterization in exhaust for three-way (TWC) [34], selective catalytic reactions reduction[32,33], catalysts (SCR) [35], of catalyst catalysts powder samples in situ during chemical or the state monitoring of catalysts directly[36]. in exhaust three-way [34], selective catalytic and diesel particulate filters (DPF) All arefor based on thecatalysts fact that(TWC) the metallic catalyst or filter (SCR) [35], and diesel particulate housingreduction form an catalysts electromagnetic waveguide cavity [37].filters (DPF) [36]. All are based on the fact that the metallic catalyst or filter housing form electromagnetic waveguide cavity [37]. Typical planar resonance structures withan microstrips are ribbon resonators, disk resonators, or Typical planar resonance structures with microstrips are ribbon resonators, disk resonators, or ring resonators [31]. Owing to their geometry, standing waves form. They are trapped between the ring resonators [31]. Owing to their geometry, standing waves form. They are trapped between the boundary conditions of the designed microstrips. Microstrip resonators must be coupled with one or boundary conditions of the designed microstrips. Microstrip resonators must be coupled with one or two feed lines excitation. Two Two types of information can bebeobtained: reflectionparameters parameters two feedfor lines for excitation. types of information can obtained: the the reflection S11S , 11 , where only one coupled feed line is required; and the reflection/transmission type. Here, two feed lines where only one coupled feed line is required; and the reflection/transmission type. Here, two feed are used, allowing evaluation the reflection S11 as wellS11asasthe transmission parameter lines are used, allowing of evaluation of theparameter reflection parameter well as the transmission S21 [31].parameter S21 [31]. The standing onstructures the structures perturbedby byaasample. sample. In The standing waveswaves on the cancan be be perturbed In the thecase caseofofgas-sensitive gas-sensitive layers, the resonance condition is then not only dependent on the permittivity of but layers, the resonance condition is then not only dependent on the permittivitytheofsubstrates, the substrates, also from the cover layers, like the here-used zeolites. Then, the effective permittivity ε eff describes but also from the cover layers, like the here-used zeolites. Then, the effective permittivity εeff describes the combined permittivities of the substrate and gas-sensitive layer. This leads to the resonant the combined permittivities of the substrate and gas-sensitive layer. This leads to the resonant frequency fres: frequency f res : c (1) f res = √ , , (1) Lch ε eff where Lch denotes the respective characteristic resonance length. It can be obtained by the design where Lch denotes the respective characteristic resonance length. It can be obtained by the design equations of Hammerstad and Jensen [31]. The characteristic resonance length depends only on the equations of Hammerstad and Jensen [31]. The characteristic resonance length depends only on the resonance geometry used. For a ring resonator, Lch = 2·π·R can be used [31]. For the meaning of resonance geometry used. For a ring resonator, Lch = 2·π·R can be used [31]. For the meaning of the the ringring resonator radius, the resonant resonantfrequency, frequency,thethe 3dB-bandwidth BWis3dB resonator radius,see seeFigure Figure1.1. Besides Besides the 3dB-bandwidth BW3dB is another characteristic parameter ofof microwave It leads leadstotothe theloaded loadedquality quality factor another characteristic parameter microwaveresonators. resonators. It factor Q, Q, which can denote a measure for the of the resonance which can denote a measure foraccuracy the accuracy of the resonancepeak peak[31]. [31]. (a)

(b)

Figure 1. Microstrip ring resonators: (a) basic setup and (b) transmission-type ring resonator with two

B d 3

f res BW3dB

s

=

Q Q=

eW fr B

Figure 1. Microstrip ring resonators: (a) basic setup and (b) transmission-type ring resonator with two capacitively coupled feed lines for excitation. capacitively coupled feed lines for excitation.

(2) (2)

All in all, theall,transducing principle can understood. With gases that adsorb All in the transducing principle canthen thenbe be easily easily understood. With gases that adsorb in the in the sensitive layer, the overall complex effective permittivity changes and a resonant frequency and sensitive layer, the overall complex effective permittivity changes and a resonant frequency and factoroccurs. shift occurs. quality quality factor shift

Sensors 2017, 17, 2422

4 of 14

3. Sensor Design Sensors 2017, 17, 2422

4 of 14

3.1. Microstrip Ring Resonator Design 3. Sensor Design

In this work, a microstrip ring resonator is used. The lack of open ends decreases the radiation losses 3.1. Microstrip Resonator Design and thus increasesRing the quality factor compared with open-end resonators. The main design parameters are the coupling gap d between the ring resonator and the theends microstrip width W of the ring In this work, a microstrip ring resonator is used. Thefeed lacklines; of open decreases the radiation (bothlosses leading the characteristic impedance); and the mean ring radius R, whichThe determines the base andtothus increases the quality factor compared with open-end resonators. main design parameters are the coupling gapthese d between the ringinfluence resonator the andelectric the feedfield lines;distribution the microstrip resonant frequency (Figure 1). All parameters atwidth resonance W ofthe thesensitivity ring (both as leading to the the resolution characteristic the meanpeak. ring To radius R, which their and thus well as of impedance); the observedand resonance understand determines the base resonant frequency (Figure 1). All these parameters influence the electric field impact, a detailed simulation study was conducted. It is discussed in the following. distribution at resonance and thus the sensitivity as well as the resolution of the observed resonance peak. To understand 3.2. Simulation Studies their impact, a detailed simulation study was conducted. It is discussed in the following.

The simulations were implemented using the RF-module of the FEM software COMSOL Multiphysics. TheStudies general structure of the model includes 50 Ω SMA connectors, two 50 Ω feed 3.2. Simulation lines, andThe thesimulations microstrip were ring resonator. the first models, RO4003C (manufactured by Rogers implementedFor using the RF-module of the FEM software COMSOL Corp.) specifications used as a substrate material (substrate thickness h = two 0.813 permittivity Multiphysics. Thewere general structure of the model includes 50 Ω SMA connectors, 50mm; Ω feed lines, ε’ = 3.38; lossmicrostrip tangent tan = ε”/ε’ = 0.0027). and the ringδ resonator. For the first models, RO4003C (manufactured by Rogers Corp.) specifications were used as a substrate material (substrate thickness h = 0.813 mm; permittivity ε’ =

3.3. Influence the Coupling Gap = 0.0027). 3.38; lossof tangent tan δ = ε’’/ε’

To study the influence of the coupling gap, the distance d between the ring resonator and the 3.3. Influence of the Coupling Gap feed lines of a transmission-type microstrip ring resonator was increased. Generally, it is known that To study the of the gap,signal the distance between the ring and thein the a larger coupling gapinfluence improves the coupling transferred since dthe electrical fieldresonator perturbation feed lines of a transmission-type microstrip ring resonator was increased. Generally, it is known that coupling gap is larger for small coupling gaps. Furthermore, it influences the resonant frequency and a larger coupling gap improves the transferred signal since the electrical field perturbation in the the resolution of the resonance peak (quality factor). To be more precise, a loose coupling yields a high coupling gap is larger for small coupling gaps. Furthermore, it influences the resonant frequency and quality and thus good accuracy (undercoupling), whereas a high acoupling level isyields needed thefactor resolution of thea resonance peak (quality factor). To be more precise, loose coupling a for power transfer (overcoupling) [31]. The required level of coupling is complex and simulation studies high quality factor and thus a good accuracy (undercoupling), whereas a high coupling level is of theneeded coupling d help to identify the influences of coupling andistocomplex estimateand needed for gap power transfer (overcoupling) [31]. The required gap levelvariations of coupling values. The parameters varied with first resonantthe frequency = 4 GHz,gap microstrip width simulation studies ofwere the coupling gap dthe help to identify influencesf res of coupling variations and to estimate needed values.of The were varied withgap the first fresand = 4 was W = 1.86 mm, and an impedance 50parameters Ω. The smallest coupling wasresonant set to d =frequency 0.025 mm GHz, in microstrip width W to = 1.86 mm, and an impedance of 50 Ω. The smallest coupling gap was set increased 0.025 mm steps 0.300 mm. to d = 0.025 mm and was increased in 0.025 to 0.300 mm. The expected behavior with increasingmm gapsteps width is confirmed by the simulations in Figure 2. The expected behavior with increasing gap width is confirmed by the simulations in Figure 2. The quality factor is higher (smaller peak width) for loosely coupled resonators and the resonant The quality factor is higher (smaller peak width) for loosely coupled resonators and the resonant frequency increases as well due to the perturbation at the coupling gap. Using the reflection parameter frequency increases as well due to the perturbation at the coupling gap. Using the reflection S11 , an increased broadens thebroadens minimum, the resonant cannot becannot as accurately parameter S11coupling , an increased coupling thei.e., minimum, i.e., thefrequency resonant frequency be determined. This phenomenon has to be taken into account if the reflection parameter is used as the as accurately determined. This phenomenon has to be taken into account if the reflection parameter microwave and the coupling gap has to begap selected is used signal, as the microwave signal, and thed coupling d has todifferently. be selected differently.

Figure 2. Scattering parameters thecoupling coupling gap gap simulation (d (d = 0.025 mmmm to 0.300 mm in Figure 2. Scattering parameters ofofthe simulationstudy study = 0.025 to 0.300 mm in steps of 0.025 mm). steps of 0.025 mm).

Sensors 2017, 17, 2422

5 of 14

3.4. Miniaturization Sensors 2017, 17, 2422

5 of 14

For miniaturized devices, the microstrip resonators have to be operated at higher frequencies. However, then the microstrip losses also increase and the quality factor gets smaller. In addition, a small 3.4. Miniaturization ratio W/R is critical due to curving effects that influence the resonator characteristics. Consequences For miniaturized devices, the microstrip resonators have to be operated at higher frequencies. of such small W/R-ratios may be parasitic resonances as well as higher losses. To account for this However, then the microstrip losses also increase and the quality factor gets smaller. In addition, a effect,small another parameter study for different resonant frequencies from 1 GHz up to 10 GHz was ratio W/R is critical due to curving effects that influence the resonator characteristics. conducted. The characteristic impedance kept to as 50 well Ω, the width was Consequences of such small W/R-ratios(ring maymicrostrip) be parasiticwas resonances as ring higher losses. Toset to W = 1.86 mm, and the coupling gap to d = 0.15 mm. The ring radius R was modified between 2.88 account for this effect, another parameter study for different resonant frequencies from 1 GHz up to mm and 29.2 mm was in a conducted. way such that frequencies between 1 GHz andwas 10 GHz occurred. Table 1, 10 GHz The resonant characteristic impedance (ring microstrip) kept to 50 Ω, theInring width wasradius set to W = 1.86and mm,the andrespective the coupling gap to d =frequencies 0.15 mm. Theare ring radius R was modified the exact ring values resonance listed. between 2.88 mm and 29.2 mm in a way such that resonant frequencies between 1 GHz and 10 GHz Table 1. Ring radius toexact obtain resonant f resrespective between 1resonance and 10 GHz. The ringare width W occurred. In Table 1, R the ring radius frequencies values and the frequencies listed. was set to 1.86 mm. Table 1. Ring radius R to obtain resonant frequencies fres between 1 and 10 GHz. The ring width W was fset to 1.86 mm. 1 2 3 4 5 6 7 8 9 10 res /GHz R/mm /GHz29.2 1 14.6 2

R/mm

29.2

14.6

9.723

9.72

7.28 4

7.28

5.81 5

5.81

4.84 6

4.84

74.14 8 3.62 9 3.2110 2.88 4.14

3.62

3.21

2.88

In Figure 3, the results of the miniaturization study are plotted. At higher resonant frequencies, In Figure 3, the results ofi.e., the miniaturization study are At higher resonant frequencies, the resonance peaks broaden, the quality factors getplotted. smaller. Furthermore, the transferred resonance peaks broaden, theto quality factors get smaller. Furthermore, signal signaltheincreases. This could bei.e., due the smaller distances that have tothe betransferred overcome. Finally, increases. This could be due to the smaller distances that have to be overcome. Finally, the the miniaturization study shows the importance of increasing the quality factor by other parameters to miniaturization study shows the importance of increasing the quality factor by other parameters to ensure a proper resolution for sensing. Hence, a miniaturization by using high-permittivity substrates ensure a proper resolution for sensing. Hence, a miniaturization by using high-permittivity should also be considered to improve the sensor signal. Despite the miniaturization improvements, substrates should also be considered to improve the sensor signal. Despite the miniaturization the microstrip substrate has othersubstrate impactshas onother the sensor These criteria and the improvements, the microstrip impactsdevice on thecharacteristics. sensor device characteristics. These here-used substrate are briefly discussed in the following. criteria and the here-used substrate are briefly discussed in the following.

Figure 3. Scattering parameters for the miniaturization simulation study. The ring radius R was varied

Figure 3. Scattering parameters for the miniaturization simulation study. The ring radius R was varied between 2.88 mm and 29.2 mm, so that the resonant frequencies increased from 1 GHz to 10 GHz. between 2.88 mm and 29.2 mm, so that the resonant frequencies increased from 1 GHz to 10 GHz.

Choosing the substrate plays a key role in the design of microstrip circuits and is widely

Choosing theliterature substrate plays a key in theto design of microstrip circuits is widely discussed discussed in [38]. Since it isrole intended heat up the device in furtherand works, an alumina substrate withSince a purity 99.6% and used. Alumina ensures asubstrate very low with in literature [38]. it is of intended to gold heat as upmetallization the device inis further works, also an alumina loss of tangent (ε’and = 10.1 @ 1 GHz, tan δ = 0.00022is@used. 1 MHz, h = 0.635 mm; sheetaofvery CeramTec Rubalit a purity 99.6% gold as metallization Alumina also data ensures low loss tangent 710). On the other hand, concerning the theory of field distribution in microstrips, the high (ε’ = 10.1 @ 1 GHz, tan δ = 0.00022 @ 1 MHz, h = 0.635 mm; data sheet of CeramTec Rubalit 710). On the permittivity negatively affects the sensitivity, since the field concentrates in the substrate with its other hand, concerning the theory of field distribution in microstrips, the high permittivity negatively higher permittivity. However, a higher permittivity allows smaller devices. affects the sensitivity, since the field concentrates in the substrate with its higher permittivity. However, a higher permittivity allows smaller devices. 3.5. Final Microwave Gas-Sensing Device Generally, aGas-Sensing design with Device only one port is desired for gas detection due to the simplified setup. 3.5. Final Microwave However, a two-port microwave transducer delivers more information than a one-port reflection-

Generally, a design only one port isadesired forand gas adetection due to the simplified type resonator. Thus,with differences between reflectiontransmission-type resonator withsetup. However, a two-port microwave transducer delivers more information than a one-port reflection-type

Sensors 2017, 17, 2422

6 of 14

resonator. Thus, differences between a reflection- and a transmission-type resonator with respect to the sensitivity were simulated. The sensitivity is here defined as a change of the resonant frequency towards a change in the intrinsic material properties of a ring covering layer: Sensors 2017, 17, 2422 6 of 14 ∆ f res is here defined as a change of the resonant f respect to the sensitivity were simulated. The Sε ressensitivity = 0 . frequency towards a change in the intrinsic material∆ε properties of a ring covering layer: eff

(3)

For the simulations, the mean radius was set to Δf R = 2.07 mm, the covering layer thickness (zeolite) Sεf res = res . (3) ' was 0.1 mm, the coupling gaps were set to 0.15 mm,Δand the Al2 O3 substrate (ε’ = 10.1) was chosen to ε eff be 0.635 mm. For the simulations, the mean radius was set to R = 2.07 mm, the covering layer thickness (zeolite) The setup of both simulated resonator types is depicted in Figure 4a,b. The complex permittivity was 0.1 mm, the coupling gaps were set to 0.15 mm, and the Al2O3 substrate (ε’ = 10.1) was chosen to of the zeolite was increased in the real and the imaginary part to account for the polarization change be 0.635 mm. and a change in intrinsic with increasing or water [35]. A permittivity list of the study The setup of both losses simulated resonator typesabsorbed is depictedammonia in Figure 4a,b. The complex steps of used is depicted in Table 2. demonstrates that, selected data, the sensitivity the zeolite was increased in Figure the real5and the imaginary partfor to the account for the polarization changeof the reflection-type device is higher and the resonance peak isammonia less broad than [35]. for the transmission-type and a change in intrinsic losses with increasing absorbed or water A list of the study steps used is depicted in Table 2. Figure 5 demonstrates that, for the selected data, the was sensitivity sensor. Consequently, in the following measurements, the reflection-type resonator used. of the reflection-type device is higher and the resonance peak is less broad than for the transmissiontype sensor. Consequently, in the following measurements, reflection-type resonator used. Table 2. Data for the sensitivity simulation. The complexthe permittivities were roughlywas estimated from [32]. Table 2. Data for the sensitivity simulation. The complex permittivities were roughly estimated from [32].

Step Step 1 1 22 ε’ 3 + 3j0+ j0 3.2 3.2++j0.1 j0.1 ε’

33 44 5 5 6 6 3.4 + j0.4 3.4 ++j0.2 j0.2 3.6 3.6+ +j0.3 j0.3 3.83.8 + j0.4 4 + j0.5 4 + j0.5

Figure 4. Three-dimensional sensor models with gas-sensitive (zeolite) cover layer (blue) for the Figure 4. Three-dimensional sensor models with gas-sensitive (zeolite) cover layer (blue) for the sensitivity simulation. (a) Transmission-type and (b) reflection-type resonator. (c) Fabricated microwave sensitivity simulation. (a) Transmission-type and (b) reflection-type resonator. (c) Fabricated ring transducer withtransducer Fe-zeolitewith as sensitive and end-mounted SMA connector microwave ring Fe-zeolitelayer as sensitive layer and end-mounted SMA(specifications: connector Al2 O(specifications: and 99.6% a thickness 0.635 mm, Rof = 0.635 2.07 mm, 0.15 mm, mm,dW = 0.22 mm). 3 with 99.6% purity Al2O3 with purity of and a thickness mm, d R= = 2.07 = 0.15 mm, W= 0.22 mm).

Finally, the coupling gap, microstrip width, and first resonance were chosen based on the Finally, the coupling gap, microstrip width, and first resonance were chosen based on the performed simulations. To miniaturize the microwave transducer, the resonant frequency of the performed simulations. To miniaturize the microwave transducer, the resonant frequency of the uncovered ring resonator was was set toset9 GHz, which leadsleads to a ring of only = 2.07 uncovered ring resonator to 9 GHz, which to a radius ring radius of R only R =mm. 2.07 However, mm. the simulation studies showed that this would lead to a smaller accuracy of the resonance However, the simulation studies showed that this would lead to a smaller accuracy of the resonancepeak. To counter this, the this, ringthe strip was set = 0.22 very small improves peak. To counter ringwidth strip width wasto setW to W = 0.22mm, mm, which which isisvery small andand improves accuracy as well as sensitivity due concentration above substrate for smaller accuracy as well as sensitivity duetotothe thehigher higher field field concentration above thethe substrate for smaller widths. value = 0.15 mmwas wasempirically empirically determined thethe preliminary fabricated ring ring strip strip widths. TheThe value d =d0.15 mm determinedbyby preliminary fabricated resonators and denotes a compromise between signal intensity and resonance peak accuracy. Finally, resonators and denotes a compromise between signal intensity and resonance peak accuracy. Finally, the designed microwave resonator alumina substratewas wascovered covered with with the the designed microwave ringring resonator onon alumina substrate the zeolite. zeolite.Due Duetoto the the mechanical instability of the zeolite cover layer, a thickness measurement was difficult. mechanical instability of the zeolite cover layer, a thickness measurement was difficult. Additionally, Additionally, the cover layer is not exactly planar, which leads to different thicknesses across the ring resonator. The thickness is assumed to be between 0.5 mm and 1 mm.

Sensors 2017, 17, 2422

7 of 14

the cover layer is not exactly planar, which leads to different thicknesses across the ring resonator. Sensors 2017,is 17,assumed 2422 7 of 14 The thickness to be between 0.5 mm and 1 mm.

Figure 5. Scattering parameters of the sensitivity simulation with changing complex permittivity from

Figure 5. Scattering parameters of the sensitivity simulation with changing complex permittivity from 3 + j0 to 4 + j0.5. Two resonator configurations were simulated: a transmission-type sensor with two 3 + j0 to 4 + j0.5. Two resonator configurations were simulated: a transmission-type sensor with two excitation lines (2-port measurement, left figure) and a reflection-type sensor with only one excitation excitation lines (2-port measurement, left figure) and a reflection-type sensor with only one excitation line (1-port measurement, right figure). The mentioned sensitivities are calculated from a linear fit of line (1-port measurement, figure). The mentioned sensitivities are calculated from a linear fit of the resonance frequencyright change. the resonance frequency change. 4. Experiment and Methods

4. Experiment Methods substrates were purchased from CeramTec (Rubalit 710, thickness 0.635 mm, ε’ = 10.1 @ 1 Al2O3 and GHz, tan δ = 0.00022 @ 1 MHz) and in-house coated by evaporation with a 400nm gold layer (adhesion Al 2 O3 substrates were purchased from CeramTec (Rubalit 710, thickness 0.635 mm, ε’ = 10.1 @ 1 GHz, layer 4 nm chromium). The microstrip ring, as well as the microstrip feed lines, were structured using tan δ = 0.00022 @ 1 MHz) and in-house coated by evaporation with a 400nm gold layer (adhesion a frequency-tripled Nd:YAG-Laser (LPKF Microline 350L, LPKF Laser & Electronics AG, Garbsen, layer 4 nm chromium). The microstrip ring, as well as the microstrip feed lines, were structured using Germany). The laser was also used to drill holes at the ends of the feed lines. They served to fix the a frequency-tripled Nd:YAG-Laser (LPKF Microline Laser & Electronics AG, Garbsen, end launch (Southwest Microwave, Tempe, AZ, USA)350L, at theLPKF ends of the substrate. Both types—the Germany). The laser was also used to drill holes at the ends of the feed lines. They served to fix the reflection type with one port and the transmission type with two ports—were fabricated this way. end launch Microwave, Tempe, USA) at the ends(PSA, of thefurther substrate. Bothintypes—the The(Southwest Fe-zeolite powder (obtained from AZ, an automotive catalyst described [39])) reflection type with port and the of transmission type with two ports—were fabricated way. was mixed with one a 1:11 compound ethyl cellulose and terpineol, which was then appliedthis on the microstrip ring and fired at 600 °C (layer thickness between 0.5 mm and 1 mm). After this, the The Fe-zeolite powder (obtained from an automotive catalyst (PSA, further described in [39])) was were attached and coaxial cables connected. The entire sensoron is the shown in mixedconnectors with a 1:11 compound of ethyl cellulose and terpineol, whichreflection-type was then applied microstrip Figure 4c. This type of zeolite was used because we know its electrical properties in the microwave ◦ ring and fired at 600 C (layer thickness between 0.5 mm and 1 mm). After this, the connectors range from microwave-based measurments in exhaust gas catalysts [39]. were attached and coaxial cables connected. The entire reflection-type sensor is shown in Figure 4c. The experimental setup included a glass sample chamber containing the microwave transducer, This type of zeolite was used because we know its electrical properties in the microwave range from a vector network analyzer (VNA: Anritsu MS2820B), a data acquisition system (Laptop), and a gas microwave-based measurments in exhaust gas catalysts supply unit connected to the sample chamber. The latter[39]. consisted of three mass flow controllers to The experimental setup included a glass sample chamber containing microwave transducer, a control the flows of N2, NH3, and the water-saturated N2. The total gas the flow rate was set to 0.5 vectormL/min. network analyzer (VNA: Anritsu MS2820B), a data acquisition system (Laptop), and a gas supply The network analyzer with connected coaxial cables was calibrated using a SOLT calibration unit connected theCalibration sample chamber. flow controllers to control kit (Anritsuto SMA Kit 3650).The Thelatter setup consisted is depictedof in three Figuremass 6. reflection parameter S11 was used. complex reflection the flowsAs of the N2 sensor , NH3 ,response, and thethe water-saturated N2 . The total gasThe flow rate was set toparameter 0.5 mL/min. S11 was measured a frequency rangecoaxial from 8 GHz to 10 GHz every 10 to 20 seconds. before kit The network analyzerinwith connected cables was calibrated using a SOLTThen, calibration the resonant frequency and the absolute value of the reflection coefficient (|S 11|) were determined, (Anritsu SMA Calibration Kit 3650). The setup is depicted in Figure 6. the reflection spectrum around the operating resonant frequency was filtered using a Lorentz fit to As the sensor response, the reflection parameter S11 was used. The complex reflection parameter remove noise and parasitic reflections. S11 was measured in a frequency range from 8 GHz to 10 GHz every 10 to 20 seconds. Then, before the Generally, after each adsorption measurement, the zeolite had to be thermally regenerated in an resonant frequency the absolute value of the reflection were determined, 11 |) external furnace.and Therefore, the coated microwave transducercoefficient without the(|S SMA connector was heated the reflection spectrum around the operating resonant frequency was filtered using a Lorentz fit tokept remove up to 120 °C over 2 min and kept at 120 °C for 10 min, then heated up to 600 °C over 25 min and noiseatand reflections. 600parasitic °C for 10 min. Finally, the transducer was cooled down to room temperature over 25 min ( 21 Generally, after each measurement, the zeolite had to be thermally °C). Furthermore, after adsorption storing in ambient air, the microwave transducer was purged inregenerated N2 for severalin an minutes to desorb stored the water on themicrowave zeolite (see transducer also next section). external furnace. Therefore, coated without the SMA connector was heated up to 120 ◦ C over 2 min and kept at 120 ◦ C for 10 min, then heated up to 600 ◦ C over 25 min and kept at 600 ◦ C for 10 min. Finally, the transducer was cooled down to room temperature over 25 min (≈21 ◦ C). Furthermore, after storing in ambient air, the microwave transducer was purged in N2 for several minutes to desorb stored water on the zeolite (see also next section).

Sensors 2017, 17, 2422 Sensors 2017, 17, 2422

8 of 14 8 of 14

Figure 6. Experimental set-up with glass sample chamber containing the microwave transducer, a

Figure 6. Experimental set-up with glass sample chamber containing the microwave transducer, vector network analyzer (VNA, Anritsu MS2820B), and a gas supply unit connected to the sample a vector network analyzer (VNA, Anritsu MS2820B), and a gas supply unit connected to the chamber. sample chamber.

5. Results and Discussion

5. Results and Discussion 5.1. Initial Tests: Proof of Concept

5.1. Initial Tests: Proof of Concept

Before the measurements in the above-described setup started, initial tests were performed to

verify the concept and the repeatability of the microwave-based monitoring of the to Before thebasic measurements in the above-described setup started, adsorption initial tests were performed zeolites at room temperature. For that purpose, the microwave sensor was connected to the VNA andof the verify the basic concept and the repeatability of the microwave-based adsorption monitoring placed in different atmospheric conditions. Apart from ambient air, a rolled-neck glass partly filled zeolites at room temperature. For that purpose, the microwave sensor was connected to the VNA and with an ammonia-water solution (25 m% NH3 in H2O) was used (setup in Figure A2). The evaporation placed in different atmospheric conditions. Apart from ambient air, a rolled-neck glass partly filled of ammonia into the gas phase established an NH3 partial pressure in the headspace of the liquid with an ammonia-water solution (25 m% NH3 in H2 O) was used (setup in Figure A2). The evaporation inside the glass, which was used for the first tests. Finally, four different cases were distinguished of ammonia the gas phase established an NH3 partial pressure in the headspace of the liquid with thisinto experiment: inside the glass, which was used for the first tests. Finally, four different cases were distinguished with 1. Sensor outside the glass (i.e., in air) without zeolite layer. this experiment:

1. 2. 3. 4.

2. Sensor inside the glass (ammonia-in-air ambience) without zeolite layer. 3. Sensor outside glass (i.e.,ininair) air)without with zeolite layer. Sensor outside thethe glass (i.e., zeolite layer. 4. Sensor inside glass (ammonia-in-air ambience) ambience) with zeolite layer. layer. Sensor inside thethe glass (ammonia-in-air without zeolite

Sensor outside glass (i.e., inthe air)glass with(inzeolite layer. of the liquid), both gaseous ammonia If the sensorthe is placed inside the headspace and water should adsorb in the zeolite and shift the resonant frequency towards lower Sensor inside the glass (ammonia-in-air ambience)sensors’ with zeolite layer. frequencies since the effective permittivity increases. For reference purposes, the experiment was also If the sensor is placed inside glass (in the headspace ofcases the liquid), both gaseous ammonia repeated without a zeolite cover the layer. In addition, the initial test were simulated with Comsol and water shouldThe adsorb in the andwas shift sensors’ resonant frequency Multiphysics. geometry forzeolite the models setthe to be the same as the fabricated ringtowards resonatorlower sensor (specifications: substrate thicknessincreases. of 0.635 mm, ring radius R = 2.07 coupling was gap also frequencies since the effective permittivity Formean reference purposes, themm, experiment d = 0.15 mm, ring microstrip width W = 0.22 mm, zeolite thickness 0.75 mm). The Al 2 O 3 substrate repeated without a zeolite cover layer. In addition, the initial test cases were simulated with Comsol properties The weregeometry chosen to befor ε’ =the 9.87 and tanwas δ = 0.0003, close toasthe corresponding datasheet Multiphysics. models set towhich be theis same the fabricated ring resonator values. Furthermore, there were no data values for the complex permittivity available for the used sensor (specifications: substrate thickness of 0.635 mm, mean ring radius R = 2.07 mm, coupling gap Fe-zeolite and the ammonia-saturated Fe-zeolite. Thus, the complex permittivities of the zeolites were d = 0.15 mm, ring microstrip width W = 0.22 mm, zeolite thickness 0.75 mm). The Al2 O3 substrate chosen in a way that they fit the measurement data best (ε’zeolite,air = 3.35, tan δzeolite,air = 0.07 and properties were chosen to be ε’ = 9.87 and tan δ = 0.0003, which is close to the corresponding datasheet ε’zeolite,NH3 = 3.97, tan δzeolite,NH3 = 0.095). These values can be thought of as a first guess of the actual values. Furthermore, valuesRoughly for the they complex available for the used permittivity of the there zeolitewere and no the data sensitivity. agreepermittivity with data from [40] for zeolites. Fe-zeolite and the ammonia-saturated Fe-zeolite. Thus, the complex permittivities of the zeolites Finally, considering the attenuation of transmission lines in the measurement setup, an attenuation were chosen in adB way they fit the measurement data best (ε’zeolite,air = 3.35, tan δzeolite,air = 0.07 and of 0.55 wasthat added to the Comsol simulation data. Figure 7 depicts the measured reflection thebefour cases. of Asasexpected, all resonance ε’zeolite,NH3 = 3.97, tan δzeolite,NH3 = 0.095). Thesespectra valuesfor can thought a first guess of the actual peaks differ from each other. frequencies without a zeolite layerdata in airfrom and in ammoniapermittivity of the zeolite and The the resonant sensitivity. Roughly they agree with [40] for zeolites. in-air are identical (① and ②). This was expected, since the gas phase above the sensor hardly of Finally, considering the attenuation of transmission lines in the measurement setup, an attenuation contributes to the resonant frequency due to their low permittivity. As soon as the microstrip ring is 0.55 dB was added to the Comsol simulation data. covered with a zeolite layer, the permittivity of the zeolite forces the fundamental resonant frequency Figure 7 depicts the measured reflection spectra for the four cases. As expected, all resonance peaks to shift to lower frequencies (③). Furthermore, by exposing the zeolite-covered sensor to the differ from each other. The resonant frequencies without a zeolite layer in air and in ammonia-in-air 1 and ). 2 are identical ( This was expected, since the gas phase above the sensor hardly contributes to the resonant frequency due to their low permittivity. As soon as the microstrip ring is covered with a zeolite layer, the permittivity of the zeolite forces the fundamental resonant frequency to shift to lower

Sensors 2017, 17, 2422 Sensors 2017, 17, 2422 Sensors 2017, 17, 2422

9 of 14

9 of 14 9 of 14

ammonia-in-air ambience, a further decrease to lower frequencies can be observed, which can be ammonia-in-air ambience, a by further decrease to lower frequencies be ammonia-in-air observed, whichambience, can be 3 frequencies ( ). Furthermore, exposing the zeolite-covered sensorcan to the attributed to the permittivity increase when ammonia and/or water adsorbs in the zeolite [41–43]. attributed to the to permittivity increase can when adsorbs in thetozeolite [41–43]. a further decrease lower frequencies beammonia observed,and/or whichwater can be attributed the permittivity From in situ experiments with zeolites applying the cavity perturbation method, the general relation From in situ experiments with zeolites applyingin the cavity perturbation method, the experiments general relation increase when ammonia and/or water adsorbs the zeolite [41–43]. From in situ with between ammonia adsorption and the electromagnetic material properties have already been studied between ammonia adsorption and the electromagnetic material properties have already beenadsorption studied zeolites applying the cavity perturbation method, the general relation between ammonia [32,35]. [32,35]. and the electromagnetic material properties have already been studied [32,35].

Figure Initial tests:absolute absolute valueof of thereflection reflection parameter |S | |ofofthe sensor forfor Figure 7. 7. Initial tests: |S11 themicrowave microwave sensor 11 Figure 7. Initial tests: absolutevalue value ofthe the reflection parameter parameter |S 11 | of the microwave sensor for different atmospheres:solid solidline line==measurement; measurement; circle line ==simulation. ① Sensor outside thethe glass 1 different atmospheres: circle line simulation. Sensor outside glass different atmospheres: solid line = measurement; circle line = simulation. ① Sensor outside the glass (i.e., in air) withoutzeolite zeolitelayer layer(black (black dashed); dashed); ② Sensor inside the glass (ammonia inin airair ambience) 2 (i.e., in air) without

Sensor inside the glass (ammonia ambience) (i.e., in air) without zeolite layer (black dashed); ② Sensor inside the glass (ammonia in air ambience) without zeolite layer(green); (green); 3③Sensor Sensor outsidethe the glass (i.e., layer (red); ④ Sensor without zeolite layer (i.e.,in inair) air)with withzeolite zeolite layer (red); Sensor without zeolite layer (green); ③ Sensoroutside outside the glass glass (i.e., in air) with zeolite layer (red); ④ 4Sensor inside the glass (ammonia in air ambience) with zeolite layer (blue). inside thethe glass (ammonia (blue). inside glass (ammoniaininair airambience) ambience)with withzeolite zeolite layer (blue).

The zeolites were regenerated and the experiment was repeated another two times. For each run, The zeolites wereregenerated regeneratedand andthe theexperiment experiment was another two times. For each run, zeolites were was repeated repeated another times. For each run, theThe resonant frequency for the unloaded case (after regeneration in air), fully two ammonia-loaded case the resonant frequency for the unloaded case (after regeneration in air), fully ammonia-loaded case the(ammonia-in-air resonant frequency for theand unloaded (after regeneration fullyexposing ammonia-loaded case ambience), now alsocase the partially loaded case in (inair), air after to ammonia) (ammonia-in-air ambience), and now also the partially loaded case (in air after exposing to ammonia) (ammonia-in-air and now alsolatter, the partially (in airwere afteragain exposing to ammonia) were recorded ambience), and compared. For the the fullyloaded loadedcase zeolites exposed to air. were recorded and compared. For the latter, the fully loaded zeolites were again exposed to air. were recorded and compared. the latter, the fully loaded zeolites were again to air. Thereby, Thereby, ammonia (definedFor as weakly bonded ammonia [44]) desorbs until aexposed new equilibrium is Thereby, ammonia (defined as weakly bonded ammonia [44]) desorbs until a new equilibrium is ammonia (defined as weakly bonded ammonia [44]) desorbs a new equilibrium is established established and only strongly bonded ammonia remains in until the zeolite. The respective resonant established and only strongly bonded ammonia remains in the zeolite. The respective resonant and only strongly bonded ammonia in the Thethe respective resonant frequencies frequencies are illustrated in Figure remains 8. It should be zeolite. noted that experiments were carried out atare frequencies are illustrated in Figure 8. It should be noted that the experiments were carried out at different in days and thus atshould different room temperatures as well as humidity. Nevertheless, in Figure illustrated Figure 8. It be noted that the experiments were carried out at different days different days and thus at different room temperatures as well as humidity. Nevertheless, in Figure 8, the resonant frequencies at the time t 0 for the empty zeolite coincide. Additionally, for all and different room temperatures as twell as humidity. Nevertheless, in Figure 8, the resonant 8, thus the at resonant frequencies at the time 0 for the empty zeolite coincide. Additionally, for all experimental runs thet0same trends canzeolite be seen in the frequency shift. The quantitative differences frequencies at the time for the empty Additionally, for all experimental runs the experimental runs the same trends can be seencoincide. in the frequency shift. The quantitative differences may be explained by the varying experimental differences in humidity and room temperature. same trends can be seen frequency shift. Thedifferences quantitative mayroom be explained by the may be explained by in thethe varying experimental in differences humidity and temperature. Furthermore, the unknown ammonia and water concentrations makeFurthermore, an explanation varying experimental differences in humidity and room temperature. thedifficult. unknown Furthermore, the unknown ammonia and water concentrations make an explanation difficult. Therefore, the subsequent tests,make the above-described sample chamber with supply unit ammonia andfor water concentrations an explanation difficult. Therefore, forthe thegas subsequent tests, Therefore, for the subsequent tests, the above-described sample chamber with the gas supply unit and defined concentrations was used. theand above-described sample chamber defined concentrations was used.with the gas supply unit and defined concentrations was used.

Figure 8. 8. Initial tests: sensorfor forthe thethree threeexperimental experimental runs Figure Initial tests:resonant resonantfrequencies frequencies of of the the microwave microwave sensor runs Figure 8. Initial tests: resonant frequencies of the microwave sensor for the three experimental runs with different loading partly loaded, loaded,und undfully fullyloaded loaded (for further with different loadingconditions conditionsof ofthe thezeolite: zeolite: empty, partly (for further with different loading conditions of the zeolite: empty, partly loaded, und fully loaded (for further details seesee text). details text). details see text).

Sensors Sensors 2017, 2017, 17, 17, 2422 2422 Sensors 2017, 17, 2422

10 10 of of 14 14 10 of 14

5.2. Tests under Defined Gas Exposure 5.2.Tests Testsunder underDefined DefinedGas GasExposure Exposure 5.2. Since humidity in air should affect the resonant frequency, magnitude and timely response Since humidity humidity in in air air should should affect affect the the resonant resonant frequency, frequency,magnitude magnitudeand andtimely timelyresponse response Since behavior were observed during water desorption. A sensor that was stored in air was installed in the behavior were observed during water desorption. A sensor that was stored in air was installed inthe the behavior were observed during6 water desorption. A sensor that was stored in air was installed in test rig as described in Figure and was then purged with pure N 2. The respective reflection spectra testrig rigas asdescribed describedininFigure Figure66and andwas wasthen thenpurged purgedwith withpure pureNN.2.The Therespective respectivereflection reflectionspectra spectra test 2 denotes around resonant frequency are shown in Figure 9.9. The blue curve aa spectrum after storage around resonant frequency are shown in Figure The blue curve denotes spectrum after storage around resonant frequency are shown inrecorded Figure 9. after The blue curve denotes a spectrum after storage in in The black and the red curve were purging with N can clearly see that both inair. air. The black and the red curve were recorded after purging with N.22.One .One Onecan canclearly clearlysee seethat thatboth both air. The black and the red curve were recorded after purging with N 2 ffres and |S11| are affected due to the incorporation of water into the zeolite during storage in ambient res and |S11| are affected due to the incorporation of water into the zeolite during storage in ambient fair. |S | are affected due to the incorporation of water into the zeolite during storage in after ambient res and 11 Since water increases the permittivity of the zeolites, the resonant frequency increases the air.Since Sincewater waterincreases increasesthe thepermittivity permittivityof ofthe thezeolites, zeolites,the theresonant resonantfrequency frequencyincreases increasesafter afterthe the air. desorption of water. The results also show that after 10 min most of the water has already desorbed, desorption of water. The results also show that after 10 min most of the water has already desorbed, desorption of water. The results also show that after 10 minitmost of the water haseach already desorbed, even at room temperature (Figure 9b). On the other hand, isis clear that before measurement, even at room temperature (Figure 9b). On the other hand, it clear that before each measurement, even at room temperature (Figure 9b). On the other hand, itreproducible is clear that before each measurement, the the has to be for minutes to results. thesensor sensor has bepurged purged forseveral several minutes toobtain obtain reproducible results. sensor has to betopurged for several minutes to obtain reproducible results.

(a) (a)

(b) (b)

Figure 9. Absolute value value of the the reflectionparameter parameter |S of the microwave sensor when stored Figure ||| of Figure 9.9. Absolute Absolute value of of the reflection reflection parameter |S |S111111 of the the microwave microwave sensor sensor when when stored stored in in in ambient air and when purged with nitrogen: (a) overall response from 8 to(b)9 near GHz (b) near ambient ambientair airand andwhen whenpurged purgedwith withnitrogen: nitrogen:(a) (a)overall overallresponse responsefrom from88to to99GHz GHz (b) nearresonance resonance resonance response. response. response.

Then, the influence of defined water content was investigated. Figure 10 isisaaatime-dependent time-dependent Then, Then,the theinfluence influenceof ofdefined definedwater watercontent contentwas wasinvestigated. investigated.Figure Figure10 10is time-dependent plot of f res and |S 11|@f res for an admixture of 1 vol % and 2 vol % of water into nitrogen. One can clearly plot of f and |S |@f for an admixture of 1 vol % and 2 vol % of water into nitrogen. One can res an admixture of 1 vol % and 2 vol % of water into nitrogen. One can plot of fres res and |S1111 |@fres for clearly see that the device responds to water. This excludes O 2 or CO 2 (both from air) from being responsible clearly the responds device responds water. This excludes frombeing air) from being see thatsee thethat device to water.toThis excludes O2 or CO2O(both from air) from responsible 2 or CO 2 (both for in 9. adsorbed water can even responsible foreffect the sensor effect in Figure 9. Again, adsorbed water canin bewater-free desorbednitrogen in water-free forthe thesensor sensor effect inFigure Figure 9.Again, Again, adsorbed water canbe bedesorbed desorbed in water-free nitrogen even at room temperature, but the process requires a longer time. Here, higher temperatures would nitrogen even at room temperature, but the process requires a longer time. Here, higher temperatures at room temperature, but the process requires a longer time. Here, higher temperatures would be be favorable. would be favorable. favorable.

Figure 10. Resonant frequency (a) and reflection parameter resonant frequency when exposed Figure parameter atat resonant frequency (b)(b) when exposed to Figure10. 10.Resonant Resonantfrequency frequency(a) (a)and andreflection reflection parameter at resonant frequency (b) when exposed to pure N 2 and to 1 vol % water or to 2 vol % water in N 2. pure N and to 1 vol % water or to 2 vol % water in N . to pure2 N2 and to 1 vol % water or to 2 vol % water in2 N2.

In further experiments, defined NH 3 concentrations were admixed. The sensor was exposed to Infurther furtherexperiments, experiments,defined definedNH NH 3 concentrations wereadmixed. admixed. The sensor was exposed In were The sensor was exposed to to a 3 concentrations aastepwise-changing NH 3 concentration from 0 ppm to 500 ppm (t1) and from 500 ppm to 1000 ppm stepwise-changing NH 3 concentration from0 0ppm ppmtoto500 500ppm ppm(t(t andfrom from500 500ppm ppmtoto1000 1000ppm ppm stepwise-changing NH from 3 concentration 1 )1)and NH 2) without regenerating the zeolite in between. Again, the obtained results in Figure 11 show NH333 (t without regenerating regenerating the the zeolite zeolite in in between. between. Again, Again, the the obtained obtained results results in inFigure Figure11 11show show NH (t(t22)) without different resonant frequency shifts during the loading with 500 ppm and 1000 ppm NH In the first differentresonant resonantfrequency frequencyshifts shiftsduring duringthe theloading loadingwith with500 500ppm ppmand and1000 1000ppm ppmNH NH33.3..In Inthe thefirst first different measurement step between t 1 and t2, when loading with 500 ppm NH3, it can be clearly seen that the measurementstep stepbetween betweent1t1and andt2t2,,when when loading loadingwith with500 500ppm ppmNH NH33,, itit can canbe beclearly clearlyseen seenthat thatthe the measurement sensor sensorsignal signalisisaccumulating accumulatinguntil untilsaturation saturationwith withammonia ammoniaisisreached. reached.Furthermore, Furthermore,increasing increasingthe the

Sensors 2017, 17, 2422

11 of 14

Sensors 2017, 17, 2422

11 of 14

sensor signal is accumulating until saturation with ammonia is reached. Furthermore, increasing the ammonia concentration to aa distinguishable distinguishable but but smaller smaller resonant resonant frequency frequency ammonia concentration to to 1000 1000 ppm ppm also also leads leads to shift. The differences in the sensor signal between the 500 ppm saturation and the 1000 ppm shift. The differences in the sensor signal between the 500 ppm saturation and the 1000 ppm saturation saturation is due to the higher partial pressure of NH 3 thatacauses a shift towards the adsorbed is due to the higher partial pressure of NH3 that causes shift towards the adsorbed phase phase in the in the adsorption isotherm. The with feed 1000 with ppm 1000 ppm ammonia is finally stopped andgas the sensor gas sensor adsorption isotherm. The feed ammonia is finally stopped and the test test chamber is purged with pure N 2 (t 3 ). Consequently, the resonant frequency shift decreases, chamber is purged with pure N2 (t3 ). Consequently, the resonant frequency shift decreases, but notbut to not initial to the value initialas value as probably expected fromdesorption. water desorption. It is assumed that only so-called the probably expected from water It is assumed that only so-called weakly weakly NH bonded NH3 can desorb at room temperature. Strongly bonded ammonia remains on the bonded 3 can desorb at room temperature. Strongly bonded ammonia remains on the zeolite even zeolite even in a nitrogen at this low temperature. These agree well with in a pure nitrogenpure atmosphere at atmosphere this low temperature. These results agree wellresults with data obtained for ◦ data obtained >200 °C for anobtained H-ZSM5inzeolite obtained in an electrical in situ cavity resonator [40], >200 C for an for H-ZSM5 zeolite an electrical in situ cavity resonator [40], if one extrapolates if one extrapolates their dataHowever, to roomwith temperature. However, applications, with respectit is toclear gas-sensing their data to room temperature. respect to gas-sensing that one applications, it is clear that one needs athe higher temperature whichto the desorption is fast enough to needs a higher temperature at which desorption is fast at enough guarantee a reversible sensor guaranteeTherefore, a reversible behavior. Therefore, it is mandatory establish a heater layer behavior. it issensor mandatory to establish a heater layer on theto sensor substrate. This will on be the the sensor substrate. Thissensor will be the next step during the sensor development. next step during the development.

Figure 11. Time dependency dependencyof ofthe theresonant resonantfrequency frequencyofofthe themicrowave microwave sensor according Figure sensor according to to Figure 4c 4c when ppm 1) or 1000 ppm NH 3 (t 2 ) are added to N 2 . At t 3 , NH 3 was turned off. when 500500 ppm (t1 )(tor 1000 ppm NH (t ) are added to N . At t , NH was turned off. 3 2 2 3 3

6. Conclusions 6. Conclusions and and Outlook Outlook This work investigated investigated the design parameters parameters of microstrip ring ring resonator resonator with with an an Fe-zeolite Fe-zeolite This work the design of aa microstrip as aa gas-sensitive gas-sensitive layer. layer. Based simulation results and commonly commonly used used microstrip microstrip design design as Based on on FEM FEM simulation results and equations, aa planar planar microstrip microstripring ringresonator resonatorstructure structureon onalumina aluminawas wasdeveloped developedand andcovered coveredwith with equations, a a Fe-zeolite film. Finally, the device was successfully operated at around 8.5 GHz as a humidity Fe-zeolite film. Finally, the device was successfully operated at around 8.5 GHz as a humidity sensor sensor for ammonia-loading detection room temperature. Theof amount stored and and forand ammonia-loading detection at room at temperature. The amount stored of water andwater ammonia ammonia in the zeolite is mirrored by both the absolute value of the reflection coefficient at resonance in the zeolite is mirrored by both the absolute value of the reflection coefficient at resonance and by andresonant by the resonant frequency itself. Due to the better resolution, we that conclude that thefrequency resonant the frequency itself. Due to the better resolution, we conclude the resonant frequency a suitable measure for and water detection thisFurthermore, planar device. is a suitableismeasure for ammonia and ammonia water detection with this planar with device. the Furthermore, the concentration-dependent frequency shifts during loading illustrate concentration-dependent frequency shifts during loading illustrate the accumulating characteristicsthe of accumulating characteristics the sensor to signal and make conceivable to concentrations measure small by amounts the sensor signal and make itofconceivable measure smallitamounts of gas using of gas concentrations bythe using the device operating in[45]. the so-called dosimeter-mode [45]. the device operating in so-called dosimeter-mode In future work, an integrated heating element will Then it may work as In future work, an integrated heating element willbebeadded addedtotothe thedevice. device. Then it may work a fast andand reversible gasgas sensor. In addition, it may help to determine adsorption isotherms. In as a fast reversible sensor. In addition, it may help to determine adsorption isotherms. combination with in situ DRIFT spectroscopy, the effect of distinct adsorbed species on the dielectric In combination with in situ DRIFT spectroscopy, the effect of distinct adsorbed species on the dielectric behavior of of the the gas gas sensitive sensitive material material can can also also be be studied studied in in aa wide wide temperature temperature regime. regime. behavior Acknowledgments: The authors would like to thank M. Dietrich and D. Rauch for the constructive discussions and The publication publication of of this this paper paper was and some some technical technical assistance. assistance. The was funded funded by by the the German German Research Research Foundation Foundation (DFG) and the University of Bayreuth in the funding program “Open Access Publishing”. (DFG) and the University of Bayreuth in the funding program “Open Access Publishing”.

Author Contributions: A.B., C.S., S.W., G.H., and R.M. conceived the experiments. J.K. manufactured the ceramic transducers of the devices in thin- and thick-film technology. A.B. performed the experiments. All together analyzed the data, evaluated the results, and wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

Sensors 2017, 17, 2422

12 of 14

Author Contributions: A.B., C.S., S.W., G.H., and R.M. conceived the experiments. J.K. manufactured the ceramic transducers of the devices in thin- and thick-film technology. A.B. performed the experiments. All together analyzed the data, evaluated the results, and wrote the paper. Sensors 2017, 2017, 17, 17, 2422 2422 Sensors Conflicts of Interest: The authors declare no conflict of interest.

12 of of 14 14 12

Appendix A A Appendix Field distribution distribution in in aa microstrip microstrip waveguide. waveguide Field waveguide

Figure A1. Microstrip waveguide: (a) cross section with ground plane and strip conductor; (b) electric Figure A1. A1. Microstrip Microstrip waveguide: waveguide: (a) (a) cross cross section section with with ground plane and strip conductor; (b) electric Figure (red) and magnetic (blue) field distribution with sample.ground plane and strip conductor; (b) electric (red) and magnetic (blue) field distribution with sample. (red) and magnetic (blue) field distribution with sample.

Simplified setup initial measurements. measurements Simplified setup for for the the initial initial measurements

Figure A2. Schematic illustration illustration of of the the simplified simplified experimental experimental setup setup for verification of of A2. Schematic Schematic illustration of the simplified experimental setup for basic basic verification Figure A2. adsorption and desorption monitoring with the microwave sensor and a VNA. monitoring with the microwave sensor and a VNA. adsorption and desorption monitoring with the microwave sensor and a VNA.

References References 1. 1. 2. 2. 3. 3. 4. 4. 5. 5.

6. 6. 7. 7.

8. 8.

9. 9.

Comini, E.; E.; Faglia, Faglia, G.; Sberveglieri, Sberveglieri, G. Solid State Gas Gas Sensing; Sensing; Springer Springer International Publishing AG: Comini, Comini, Faglia, G.; Sberveglieri, G. G. Solid Solid State Springer International International Publishing Publishing AG: AG: Heidelberg, Germany, Germany, 2009; doi:10.1007/978-0-387-09665-0. Heidelberg, 2009; doi:10.1007/978-0-387-09665-0. Germany, 2009. Bănică, F.-G. F.-G. Chemical Chemical Sensors Sensors and and Biosensors: Biosensors: Fundamentals Fundamentals and Applications; John Wiley & Sons, Ltd.: Bănică, Biosensors: Fundamentals and and Applications; Applications; John John Wiley Wiley & & Sons, Sons, Ltd.: Ltd.: Chichester, UK, 2012; pp. 246–257, doi:10.1002/9781118354162. Chichester, UK, 2012; pp. 246–257. 246–257, [CrossRef] doi:10.1002/9781118354162. Yamazoe, N.; N.; Shimanoe, Shimanoe, K. K. Fundamentals Fundamentals of of Semiconductor Semiconductor Gas Gas Sensors. Sensors. In In Semiconductor Gas Sensors; Yamazoe, In Semiconductor Semiconductor Gas Gas Sensors; Sensors; Jaaniso, R., R., Tan, Tan, O.K., O.K., Eds.; Eds.; Woodhead WoodheadPublishing PublishingLtd.: Ltd.:Cambridge, Cambridge,UK, UK,2013; 2013;pp. pp.3–34. 3–34. Jaaniso, O.K., Eds.; Woodhead Publishing Ltd.: Cambridge, UK, 2013; pp. 3–34. Barsan, N.; N.; Koziej, Koziej, D.; D.; Weimar, Weimar, U. Metal Metal Oxide-Based Oxide-Based Gas Gas Sensor Sensor Research: Research: How How to? to? Sens. Sens. Actuators Actuators B Chem. Barsan, U. Oxide-Based Gas Sensor Research: Weimar, U. Actuators B B Chem. Chem. 2007, 121, 18–35, doi:10.1016/j.snb.2006.09.047. 2007, 121, 18–35, 18–35. doi:10.1016/j.snb.2006.09.047. [CrossRef] Zarifi, M.H.; M.H.; Shariaty, Shariaty, P.; P.; Hashisho, Hashisho, Z.; Z.; Daneshmand, Daneshmand, M. M. A A Non-Contact Non-Contact Microwave Microwave Sensor Sensor for for Monitoring Monitoring Zarifi, P.; Hashisho, Z.; Daneshmand, and CH CH44 in in Gaseous Gaseous Streams. Streams. Sens. Sens. Actuators Actuators B Chem. 2017, 238, the Interaction Interaction of of Zeolite Zeolite 13X 13X with with CO CO22 and the CO Actuators B B Chem. Chem. 2017, 2017, 238, 238, 2 4 1240–1247, doi:10.1016/j.snb.2016.09.047. 1240–1247. [CrossRef] 1240–1247, doi:10.1016/j.snb.2016.09.047. Zarifi, M.H.; M.H.; Sohrabi, Sohrabi, A.; Shaibani, P.M.; Daneshmand, M.; Thundat, T. Detection of Volatile Organic Organic Sohrabi, A.; A.; Shaibani, Shaibani, P.M.; P.M.; Daneshmand, Daneshmand, M.; M.; Thundat, Thundat, T. T. Detection Detection of of Volatile Zarifi, Compounds Using Microwave Sensors. IEEE Sens. J. 2015, 15, 248–254, doi:10.1109/JSEN.2014.2345477. 248–254. doi:10.1109/JSEN.2014.2345477. [CrossRef] Compounds Using Microwave Sensors. IEEE Sens. J. 2015, 15, 248–254, Rossignol, J.; J.; Barochi, Barochi, G.; G.; de de Fonseca, Fonseca, B.; Brunet, J.; Bouvet, M.; Pauly, A.; Markey, L. Microwave-based Microwave-based gas gas Fonseca, B.; B.; Brunet, Brunet, J.; J.; Bouvet, Bouvet, M.; M.; Pauly, Pauly, A.; A.; Markey, Markey, L. Rossignol, sensor with phthalocyanine film at room temperature. Sens. Actuators B Chem. 2013, 189, 213–216, film at room temperature. Sens. Actuators B Chem. B2013, 189,2013, 213–216. sensor with withphthalocyanine phthalocyanine film at room temperature. Sens. Actuators Chem. 189,[CrossRef] 213–216, doi:10.1016/j.snb.2013.03.092. doi:10.1016/j.snb.2013.03.092. Bailly, G.; G.; Harrabi, Harrabi, A.; A.; Rossignol, Rossignol, J.; J.; Stuerga, Stuerga, D.; D.; Pribetich, Pribetich, P. P. Microwave Microwave Gas Gas Sensing Sensing with with aa Microstrip Microstrip Bailly, with TiO TiO22 nanoparticles. nanoparticles. Sens. Sens. Actuators Actuators B B Chem. Chem. 2016, 2016, 236, 236, 554– 554– Interdigital Capacitor: Capacitor: Detection Detection of of NH NH33 with Interdigital 564, doi:10.1016/j.snb.2016.06.048. 564, doi:10.1016/j.snb.2016.06.048. Bailly, G.; G.; Rossignol, Rossignol, J.; J.; de de Fonseca, Fonseca, B.; B.; Pribetich, Pribetich, P.; P.; Stuerga, Stuerga, D. D. Microwave Microwave Gas Gas Sensing Sensing with with Hematite: Hematite: Bailly, Shape Effect on Ammonia Detection Using Pseudocubic, Rhombohedral, and Spindlelike Particles. ACS

Sensors 2017, 17, 2422

8.

9.

10. 11. 12. 13. 14. 15.

16. 17. 18. 19.

20. 21.

22.

23. 24. 25.

26.

27.

28.

13 of 14

Bailly, G.; Harrabi, A.; Rossignol, J.; Stuerga, D.; Pribetich, P. Microwave Gas Sensing with a Microstrip Interdigital Capacitor: Detection of NH3 with TiO2 nanoparticles. Sens. Actuators B Chem. 2016, 236, 554–564. [CrossRef] Bailly, G.; Rossignol, J.; de Fonseca, B.; Pribetich, P.; Stuerga, D. Microwave Gas Sensing with Hematite: Shape Effect on Ammonia Detection Using Pseudocubic, Rhombohedral, and Spindlelike Particles. ACS Sens. 2016, 1, 656–662. [CrossRef] Chen, L.F.; Ong, C.K.; Neo, C.P.; Varadan, V.V.; Varadan, V.K. Microwave Electronics: Measurement and Materials Characterization; John Wiley & Sons, Ltd.: Chichester, UK, 2004. Yeow, Y.; Abbas, Z.; Khalid, K. Application of Microwave Moisture Sensor for Determination of Oil Palm Fruit Ripeness. Meas. Sci. Rev. 2010, 10, 714. [CrossRef] Yogi, R.A.; Parolia, R.S.; Karekar, R.N.; Aiyer, R.C. Microwave Microstrip Ring Resonator as a Paper Moisture Sensor: Study with Different Grammage. Meas. Sci. Technol. 2002, 13, 1558–1562. [CrossRef] Sarabandi, K.; Li, E.S. Microstrip Ring Resonator for Soil Moisture Measurements. IEEE Trans. Geosci. Remote Sens. 1997, 35, 1223–1231. [CrossRef] Abegaonkar, M.P.; Karekar, R.N.; Aiyer, R.C. A Microwave Microstrip Ring Resonator as a Moisture Sensor for Biomaterials: Application to Wheat Grains. Meas. Sci. Technol. 1999, 10, 195–200. [CrossRef] Schwerthoeffer, U.; Weigel, R.; Kissinger, D. A Highly Sensitive Glucose Biosensor Based on a Microstrip Ring Resonator. In Proceedings of the IEEE MTT-S International Microwave Workshop Series on RF and Wireless Technologies for Biomedical and Healthcare Applications (IMWS-BIO), Singapore, 9–11 December 2013; pp. 1–3. [CrossRef] Zarifi, M.H.; Thundat, T.; Daneshmand, M. High resolution microwave microstrip resonator for sensing applications. Sens. Actuators A Phys. 2015, 233, 224–230. [CrossRef] Yi, F.-Y.; Chen, D.; Wu, M.-K.; Han, L.; Jiang, H.-L. Chemical Sensors Based on Metal–Organic Frameworks. ChemPlusChem 2016, 81, 675–690. [CrossRef] Marr, I.; Reiß, S.; Hagen, G.; Moos, R. Planar Zeolite film-Based Potentiometric Gas Sensors Manufactured by a Combined Thick-film and Electroplating Technique. Sensors 2011, 11, 7736–7748. [CrossRef] [PubMed] Chen, P.; Schönebaum, S.; Simons, T.; Rauch, D.; Dietrich, M.; Moos, R.; Simon, U. Correlating the Integral Sensing Properties of Zeolites with Molecular Processes by Combining Broadband Impedance and DRIFT Spectroscopy—A New Approach for Bridging the Scales. Sensors 2015, 15, 28915–28941. [CrossRef] [PubMed] Xu, X.; Wang, J.; Long, Y. Zeolite-based Materials for Gas Sensors. Sensors 2006, 6, 1751–1764. [CrossRef] Bahoumina, P.; Hallil, H.; Lachaud, J.L.; Rebiere, D.; Dejous, C.; Abdelghani, A.; Frigui, K.; Bila, S.; Baillargeat, D.; Zhang, Q.; et al. Chemical Gas Sensor Based on a Novel Capacitive Microwave Flexible Transducer and Composite Polymer Carbon Nanomaterials. In Proceedings of the Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS (DTIP), Cannes, France, 1–4 April 2017. [CrossRef] Zarifi, M.H.; Gholidoustb, A.; Abdolrazzaghic, M.; Shariatyb, P.; Hashisho, Z.; Daneshmand, M. Sensitivity Enhancement in Planar Microwave Active-Resonator Using Metal Organic Framework for CO2 Detection. Sens. Actuators B Chem. 2017, in press. [CrossRef] Bailly, G.; Harrabi, A.; Rossignol, J.; Michel, M.; Stuerga, D.; Pribetich, P. Microstrip Spiral Resonator For Microwave-Based Gas Sensing. IEEE Sens. Lett. 2017, 1. [CrossRef] Piekarz, I.; Socki, J.; Wincza, K.; Gruszczynski, S. Microwave Sensors for Dielectric Sample Measurement Based on Coupled-Line Section. IEEE Trans. Microw. Theory Tech. 2017, 65, 1615–1631. [CrossRef] Yakhlef, Y.; Benhabiles, M.T.; Benkhaoua, L.; Riabi, M.L. Compact Miniature Sensors Based on Tapered Lines Coupled Metamaterial Resonators. In Proceedings of the First IEEE MTT-S International Microwave Bio Conference (IMBIOC), Gothenburg, Sweden, 15–17 May 2017; pp. 1–4. Denoual, M.; Pouliquen, M.; Jorel, C.; Radu, C.; Robbes, D.; Harnois, M.; de Sagazan, O.; Grand, J.; Awala, H.; Mintova, S.; et al. Zeolite-Based Thermal Mass Gas Sensor with Self-Identification Algorithm. In Proceedings of the Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS (DTIP), Gothenburg, Sweden, 19–21 September 2017; pp. 1–4. Reinecke, T.; Walter, J.-G.; Kobelt, T.; Ahrens, A.; Scheper, T.; Zimmermann, S. Biosensor Based on a Split-Ring Resonator. In Proceedings of the SENSOR 2017, Nuremberg, Germany, 30 May–1 June 2017; pp. 78–83. [CrossRef] Rydosz, A.; Maciak, E.; Wincza, K.; Gruszczynski, S. Microwave-Based Sensors with Phthalocyanine Films for Acetone, Ethanol and Methanol Detection. Sens. Actuators B Chem. 2016, 237, 876–886. [CrossRef]

Sensors 2017, 17, 2422

29.

30.

31. 32.

33.

34.

35.

36. 37. 38. 39. 40.

41. 42. 43. 44. 45.

14 of 14

Kröcher, O.; Devadas, M.; Elsener, M.; Wokaun, A.; Söger, N.; Pfeifer, M.; Demel, Y.; Mussmann, L. Investigation of the Selective Catalytic Reduction of NO by NH3 on Fe-ZSM5 Monolith Catalysts. Appl. Catal. B Environ. 2006, 66, 208–216. [CrossRef] Reiß, S.; Schönauer, D.; Hagen, G.; Fischerauer, G.; Moos, R. Monitoring the Ammonia Loading of Zeolite-Based Ammonia SCR Catalysts by a Microwave Method. Chem. Eng. Technol. 2011, 34, 791–796. [CrossRef] Pozar, D.M. Microwave Engineering, 4th ed.; John Wiley & Sons, Ltd.: Hoboken, NJ, USA, 2012. Dietrich, M.; Rauch, D.; Porch, A.; Moos, R. A Laboratory Test Setup for in Situ Measurements of the Dielectric Properties of Catalyst Powder Samples under Reaction Conditions by Microwave Cavity Perturbation: Set up and Initial Tests. Sensors 2014, 14, 16856–16868. [CrossRef] [PubMed] Heine, C.; Girgsdies, F.; Truschke, A.; Schlögl, R.; Eichelbaum, M. The Model Oxidation Catalyst α-V2 O5 : Insights from Contactless in Situ Microwave Permittivity and Conductivity Measurements. Appl. Phys. A 2013, 112, 289–296. [CrossRef] Beulertz, G.; Votsmeier, M.; Moos, R. Effect of Propene, Propane, and Methane on Conversion and Oxidation State of Three-Way Catalysts: A Microwave Cavity Perturbation Study. Appl. Catal. B 2015, 165, 369–377. [CrossRef] Dietrich, M.; Rauch, D.; Simon, U.; Porch, A.; Moos, R. Ammonia Storage Studies on H-ZSM-5 Zeolites by Microwave Cavity Perturbation: Correlation of Dielectric Properties with Ammonia Storage. J. Sens. Sens. Syst. 2015, 4, 263–269. [CrossRef] Feulner, M.; Seufert, F.; Müller, A.; Hagen, G.; Moos, R. Influencing Parameters on the Microwave-Based Soot Load Determination of Diesel Particulate Filters. Top. Catal. 2017, 60, 374–380. [CrossRef] Moos, R. Microwave-Based Catalyst State Diagnosis—State of the Art and Future Perspectives. SAE Int. J. Engines 2015, 8, 1240–1245. [CrossRef] Hoffmann, R.K. Integrierte Mikrowellenschaltungen; Springer: Berlin/Heidelberg, Germany, 1983. Dietrich, M.; Steiner, C.; Hagen, G.; Moos, R. Radio-Frequency-Based Urea Dosing Control for Diesel Engines with Ammonia SCR Catalysts. SAE Int. J. Engines 2017, 10, 1638–1645. [CrossRef] Rauch, D.; Dietrich, M.; Simons, T.; Simon, U.; Porch, A.; Moos, R. Microwave Cavity Perturbation Studies on H-form and Cu Ion-Exchanged SCR Catalyst Materials: Correlation of Ammonia Storage and Dielectric Properties. Top. Catal. 2017, 60, 243–249. [CrossRef] Simon, U.; Flesch, U.; Maunz, W.; Müller, R.; Plog, C. The Effect of NH3 on the Ionic Conductivity of Dehydrated Zeolites Na Beta and H Beta. Microporous Mesoporous Mater. 1998, 21, 111–116. [CrossRef] Anderson, P.A.; Armstrong, A.R.; Porch, A.; Edwards, P.P.; Woodall, L.J. Structure and Electronic Properties of Potassium-Loaded Zeolite L. J. Phys. Chem. B 1997, 101, 9892–9900. [CrossRef] Kraus, M.; Kopinke, F.-D.; Roland, U. Influence of Moisture Content and Temperature on the Dielectric Permittivity of Zeolite NaY. Phys. Chem. Chem. Phys. 2011, 13, 4119–4125. [CrossRef] [PubMed] Niwa, M.; Iwamoto, M.; Segawa, K. Temperature-Programmed Desorption of Ammonia on Zeolites. Influence of the Experimental Conditions on the Acidity Measurement. BCSJ 1986, 59, 3735–3739. [CrossRef] Marr, I.; Groß, A.; Moos, R. Overview on Conductometric Solid-State Gas Dosimeters. J. Sens. Sens. Syst. 2014, 3, 29–46. [CrossRef] © 2017 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/).