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;
Functional.Materials@uni-bayreuth.de (A.B.); Functional.Materials@uni-bayreuth.de (C.S.);
Functional.Materials@uni-bayreuth.de (S.W.); Functional.Materials@uni-bayreuth.de (J.K.);
Functional.Materials@uni-bayreuth.de (G.H.)
* Correspondence: Functional.Materials@uni-bayreuth.de; 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

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
Like a result of the single plane
microwave-based structure,
material they can be fabricated
characterization methodsconventionally by thick or
based on microstrips, thin film
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, non-
due 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 due to ontheir
the resonant frequency
higher sensitivity andchange
accuracy.of the scatteringinparameters
Well-known this respect by introducing
is the resonant a
sample perturbation
[10]. Examples method, which is based on the resonant frequency change of the scattering parameters of
for such applications in the field of chemical sensors are the characterization
catalystbypowder samples
introducing a sample in situ during
[10]. chemical
Examples reactions
for such [32,33],
applications or the
in the fieldstate monitoring
of chemical sensors of are
catalysts
the
directlycharacterization
in exhaust for three-way
of catalyst catalysts (TWC) [34],
powder samples in situselective catalytic reactions
during chemical reduction[32,33],
catalysts (SCR)
or the [35],
state
monitoring
and diesel particulateof catalysts directly[36].
filters (DPF) in exhaust
All arefor three-way
based on thecatalysts
fact that(TWC) [34], selective
the metallic catalyst catalytic
or filter
housingreduction
form an catalysts (SCR) [35],
electromagnetic and diesel particulate
waveguide cavity [37].filters (DPF) [36]. All are based on the fact that
Typical planar resonance structures withan
the metallic catalyst or filter housing form electromagnetic
microstrips waveguide
are ribbon cavity [37].
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 feedtwo lines
feedfor excitation.
lines Two Two
for excitation. types of information
types of information cancanbebeobtained:
obtained: the reflectionparameters
the reflection parameters S11S
, 11 ,
where only one coupled feed line is required; and the reflection/transmission
where only one coupled feed line is required; and the reflection/transmission type. Here, two feedtype. Here, two feed lines
are used,
linesallowing
are used, evaluation
allowing of the reflection
evaluation of theparameter S11 as wellS11asasthe
reflection parameter transmission
well parameter
as the transmission
S21 [31].parameter S21 [31].
The standing
The standing waveswaves on the onstructures
the structurescancanbe be perturbedby
perturbed byaasample.
sample. InIn the
thecase
caseofofgas-sensitive
gas-sensitive
layers, the resonance condition is then not only dependent on the permittivitytheofsubstrates,
layers, the resonance condition is then not only dependent on the permittivity of the substrates,but
also from the cover layers, like the here-used zeolites. Then, the effective permittivity
but also from the cover layers, like the here-used zeolites. Then, the effective permittivity εeff describes ε eff 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:
c

frequency f res :
fr
=

c
Lc

f res = √ , , (1)
e
s

Lch ε eff (1)

ε
h

e
f
f

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,
resonator radius,see
seeFigure
Figure1.1. Besides the resonant
Besides the resonantfrequency,
frequency,thethe 3dB-bandwidth
3dB-bandwidth BWis3dB
BW3dB
is another characteristic
another parameter
characteristic parameter ofof
microwave
microwaveresonators. It leads
resonators. It leadstotothe
theloaded
loadedquality
quality Q, Q,
factor
factor
which can denote
which a measure
can denote for the
a measure foraccuracy of the
the accuracy resonance
of the resonancepeak
peak[31].
[31].

(a) (b)

Figure 1. Microstrip ring resonators: (a) basic setup and (b) transmission-type ring resonator with two
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.
fr
Q
=
eW
s
B

f res
Q= (2) (2)
3
d
B

BW3dB
All in all, theall,transducing
All in principle
the transducing can
principle canthen
thenbe
be easily understood.
easily understood. With
With gases
gases that adsorb
that adsorb in the in
the sensitive layer, the overall complex effective permittivity changes and a resonant frequency
sensitive layer, the overall complex effective permittivity changes and a resonant frequency and and
quality quality
factor shift
factoroccurs.
shift occurs.
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
and thus increasesRing Resonator
the quality Design
factor compared with open-end resonators. The main design parameters
are the coupling gap d between the ring resonator
In this work, a microstrip ring resonator and the
is used. Thefeed
lacklines; theends
of open microstrip width
decreases W of the ring
the radiation
(bothlosses
leadingandtothus
the characteristic impedance);
increases the quality and the mean
factor compared ring radius
with open-end R, whichThe
resonators. determines
main design the base
resonant frequency
parameters (Figure
are the 1). All
coupling gapthese parameters
d between the ringinfluence
resonator the
andelectric
the feedfield
lines;distribution
the microstripatwidth
resonance
W ofthe
and thus thesensitivity
ring (both asleading
well asto the
the resolution
characteristic
of impedance);
the observedand the meanpeak.
resonance ring To
radius R, which their
understand
determines the base resonant frequency (Figure 1). All these parameters
impact, a detailed simulation study was conducted. It is discussed in the following. influence the electric field
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. general structure of the model includes 50 Ω SMA connectors, two 50 Ω feed
TheStudies
3.2. Simulation
lines, andThe thesimulations
microstrip were
ring resonator.
implementedFor the first
using models, RO4003C
the RF-module (manufactured
of the FEM by Rogers
software COMSOL
Corp.) specifications
Multiphysics. Thewere used
general as a substrate
structure material
of the model (substrate
includes thickness
50 Ω SMA h = two
connectors, 0.81350mm; permittivity
Ω feed lines,
ε’ = 3.38; lossmicrostrip
and the tangent tan = ε”/ε’ = 0.0027).
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
3.38; lossof the Coupling
tangent Gap = 0.0027).
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
a larger coupling gapinfluence
improves of the
the coupling gap,signal
transferred the distance
since dthe between the ring
electrical fieldresonator and thein the
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
qualitythefactor and thus
resolution of thea resonance
good accuracy (undercoupling),
peak (quality factor). To be whereas a high acoupling
more precise, level isyields
loose coupling needed a for
power transfer (overcoupling) [31]. The required level of coupling is complex
high quality factor and thus a good accuracy (undercoupling), whereas a high coupling level is and simulation studies
of theneeded
coupling for gap d help
power to identify
transfer the influences
(overcoupling) of coupling
[31]. The required gap levelvariations
of couplingandistocomplex
estimateand needed
values. The parameters
simulation studies ofwere varied with
the coupling gap dthe first
help resonantthe
to identify influencesf res
frequency = 4 GHz,gap
of coupling microstrip
variations width
and to
W = 1.86 estimate
mm, and an needed values.of
impedance The Ω. The smallest
50parameters were varied
couplingwithgapthe first
wasresonant
set to d =frequency
0.025 mm fresand
= 4 was
GHz, in
increased microstrip
0.025 mm width
steps W to
= 1.86 mm,
0.300 mm. and an impedance of 50 Ω. The smallest coupling gap was set
The expected behavior with increasingmm
to d = 0.025 mm and was increased in 0.025 gapsteps
widthto 0.300 mm.
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
parameter S11coupling broadens
, an increased thebroadens
coupling minimum, thei.e., the resonant
minimum, i.e., thefrequency cannot becannot
resonant frequency as accurately
be
determined. This phenomenon has to be taken into account if the reflection
as accurately determined. This phenomenon has to be taken into account if the reflection parameter parameter is used as the
microwave
is used signal, and the coupling
as the microwave signal, andgapthed coupling
has to begap selected
d has todifferently.
be selected differently.

Figure
Figure 2. Scattering
2. Scattering parameters
parameters ofofthe
thecoupling
coupling gap
gap simulation
simulationstudy (d (d
study = 0.025 mmmm
= 0.025 to 0.300 mm in
to 0.300 mm in
steps of 0.025
steps of 0.025 mm). 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
Consequences impedance
of such small W/R-ratios(ring
maymicrostrip)
be parasiticwas kept to as
resonances Ω, the
50 well as ring width
higher was
losses. Toset to
W = 1.86 and the coupling gap to d = 0.15 mm. The ring radius R
account for this effect, another parameter study for different resonant frequencies from 1 GHz up to mm
mm, was modified between 2.88
and 29.2 mm was
10 GHz in a conducted.
way such thatThe resonant frequencies
characteristic between
impedance 1 GHz andwas
(ring microstrip) 10 GHz occurred.
kept to 50 Ω, theInring
Table 1,
width
the exact wasradius
ring set to W = 1.86and
values mm,theandrespective
the coupling gap to d =frequencies
resonance 0.15 mm. Theare
ring radius R was modified
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
occurred. In Table 1, R
radius toexact
the obtain resonant
ring radius frequencies f resrespective
values and the between 1resonance
and 10 GHz. The ringare
frequencies width W
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
res /GHz 2 3 4 5 6 7 8 9 10
R/mm /GHz29.2 1 14.6 2 9.723 7.28
4 5.81
5 4.84
6 74.14 8 3.62 9 3.2110 2.88
R/mm 29.2 14.6 9.72 7.28 5.81 4.84 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
the resonance 3, the
peaks results ofi.e.,
broaden, the miniaturization study are
the quality factors getplotted.
smaller.At higher resonant frequencies,
Furthermore, the transferred
signaltheincreases.
resonance peaks broaden,
This could bei.e.,
due theto
quality factors get
the smaller smaller. Furthermore,
distances that have tothe betransferred
overcome. signal
Finally,
increases. This could be due to the smaller distances that have to be overcome.
the miniaturization study shows the importance of increasing the quality factor by other parameters toFinally, the
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
improvements,substrate has othersubstrate
the microstrip impactshas onother
the sensor
impactsdevice
on thecharacteristics. These criteria
sensor device characteristics. and the
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
discussed intheliterature
substrate plays
[38]. a key
Since it isrole in theto
intended design
heat upof microstrip circuits
the device in furtherand is widely
works, discussed
an alumina
substrate
in literature withSince
[38]. a purity
it is of 99.6% and
intended to gold
heat as
upmetallization
the device inis further
used. Alumina
works, also ensures asubstrate
an alumina very low with
loss of
a purity tangent
99.6% (ε’and
= 10.1 @ 1 GHz,
gold tan δ = 0.00022is@used.
as metallization 1 MHz, h = 0.635 mm;
Alumina also data sheetaofvery
ensures CeramTec Rubalit
low loss tangent
710). On the other hand, concerning the theory of field distribution in microstrips,
(ε’ = 10.1 @ 1 GHz, tan δ = 0.00022 @ 1 MHz, h = 0.635 mm; data sheet of CeramTec Rubalit 710). On the the high
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
3.5. Final Microwave allows smaller
Gas-Sensing devices.
Device
Generally, aGas-Sensing
3.5. Final Microwave design with Device
only one port is desired for gas detection due to the simplified setup.
However, a two-port microwave transducer delivers more information than a one-port reflection-
Generally, a design
type resonator. Thus,with only one
differences port isadesired
between forand
reflection- gas adetection due to the
transmission-type simplified
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
respect to the sensitivity were simulated. The ∆ f res is here defined as a change of the resonant
Sε ressensitivity
= 0 . (3)
frequency towards a change in the intrinsic material∆ε properties
eff of a ring covering layer:

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 ε eff
' the Al2 O3 substrate (ε’ = 10.1) was chosen to
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
The setup of both losses with
simulated increasing
resonator typesabsorbed
is depictedammonia or water
in Figure 4a,b. [35]. A permittivity
The complex list of the study
steps of
used is depicted
the zeolite in Table 2.
was increased in Figure
the real5and
demonstrates
the imaginary that,
partfor
to the selected
account data,
for the the sensitivity
polarization changeof the
reflection-type
and a change device is higher
in intrinsic and
losses withthe resonance
increasing peak isammonia
absorbed less broad than [35].
or water for the transmission-type
A list of the study
sensor. Consequently,
steps in the
used is depicted following
in Table measurements,
2. Figure 5 demonstrates the reflection-type
that, for the selectedresonator
data, the was used. of
sensitivity
the reflection-type device is higher and the resonance peak is less broad than for the transmission-
type sensor.
Table 2. DataConsequently, in the following
for the sensitivity measurements,
simulation. The complexthe reflection-type
permittivities resonator
were roughlywas used.
estimated
from [32].
Table 2. Data for the sensitivity simulation. The complex permittivities were roughly estimated from [32].
Step Step 1 1 22 33 44 5 5 6 6
ε’ ε’ 3 + 3j0+ j0 3.2
3.2++j0.1
j0.1 3.4
3.4 ++j0.2
j0.2 3.6
3.6+ +j0.3
j0.3 3.83.8
+ j0.4
+ 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
microwave ring Fe-zeolitewith
as sensitive
Fe-zeolitelayer and end-mounted
as sensitive SMA connector
layer and end-mounted SMA(specifications:
connector
Al2 O(specifications:
3 with 99.6% purity and 99.6%
Al2O3 with a thickness
purity of
and0.635 mm, Rof
a thickness = 0.635 mm, d
2.07 mm, R= mm,dW
0.15 mm,
= 2.07 = 0.22
= 0.15 mm,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
uncovered was was
ring resonator set toset9 GHz, which
to 9 GHz, leadsleads
which to a ring
to a radius
ring radius of R
of only = 2.07
only R =mm.
2.07 However,
mm.
the simulation
However, the simulation studies showed that this would lead to a smaller accuracy of the resonancepeak.
studies showed that this would lead to a smaller accuracy of the resonance
To counter
peak. To this, the this,
counter ringthe
strip
ringwidth was set
strip width wastosetW = 0.22
to W = 0.22mm,
mm, which
which isisvery
very small
small andand improves
improves
accuracy as well
accuracy as sensitivity
as well as sensitivitydue
duetotothethehigher
higher field concentration
field concentration above
above thethe substrate
substrate for smaller
for smaller
strip strip widths.
widths. TheThe value
value d =d0.15
= 0.15
mm mmwas wasempirically
empirically determined
determinedbyby thethe
preliminary
preliminary fabricated ring ring
fabricated
resonators and denotes a compromise between signal intensity and resonance
resonators and denotes a compromise between signal intensity and resonance peak accuracy. Finally,peak accuracy. Finally,
the designed
the designed microwave
microwave ringring resonator
resonator onon alumina
alumina substratewas
substrate wascovered
covered with
with the
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.
The thickness
Sensors 2017,is
17,assumed
2422 to be between 0.5 mm and 1 mm. 7 of 14

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,
the resonance frequencyright figure). The mentioned sensitivities are calculated from a linear fit of
change.
the resonance frequency change.
4. Experiment and Methods
4. Experiment
Al2O3 and Methods
substrates were purchased from CeramTec (Rubalit 710, thickness 0.635 mm, ε’ = 10.1 @ 1
GHz,
Al tan δ = 0.00022 @ 1 MHz) and in-house coated by evaporation with a 400nm gold layer (adhesion
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
end launch (SouthwestNd:YAG-Laser
Microwave, (LPKF
Tempe, Microline
AZ, USA)350L,
at theLPKF
ends of Laser & Electronics
the substrate. AG, Garbsen,
Both types—the
Germany). The laser was also used to drill holes at the ends of the feed
reflection type with one port and the transmission type with two ports—were fabricated this way. lines. They served to fix the
end launch The(Southwest Microwave,
Fe-zeolite powder Tempe,
(obtained from AZ, USA) at the
an automotive ends(PSA,
catalyst of thefurther
substrate. Bothintypes—the
described [39]))
reflection type with
was mixed with one port
a 1:11 and the of
compound transmission
ethyl cellulosetypeandwith two ports—were
terpineol, which was then fabricated
appliedthis way.
on the
microstrip ring and fired at 600 °C (layer thickness between 0.5 mm and
The Fe-zeolite powder (obtained from an automotive catalyst (PSA, further described in [39])) was 1 mm). After this, the
mixedconnectors
with a 1:11 were attached and
compound coaxial
of ethyl cables connected.
cellulose The entire
and terpineol, whichreflection-type
was then applied sensoronis the
shown in
microstrip
Figure 4c. This type◦ of zeolite was used because we know its electrical
ring and fired at 600 C (layer thickness between 0.5 mm and 1 mm). After this, the connectors properties in the microwave
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
supply unit connected in exhaust
to the sample chamber.gas catalysts
The latter[39].
consisted of three mass flow controllers to
The experimental setup included a glass sample
control the flows of N2, NH3, and the water-saturated N2. The chamber containing
total gas the flow microwave
rate was set transducer,
to 0.5 a
vectormL/min.
network analyzer (VNA: Anritsu MS2820B), a data acquisition system
The network analyzer with connected coaxial cables was calibrated using a SOLT calibration(Laptop), and a gas supply
unit connected
kit (Anritsuto SMAtheCalibration
sample chamber.Kit 3650).The Thelatter
setup consisted
is depictedof in three
Figuremass6. flow controllers to control
the flowsAs of the
N2 sensor
, NH3 ,response,
and thethe reflection parameter
water-saturated S11 was
N2 . The used.
total gasThe complex
flow rate wasreflection
set toparameter
0.5 mL/min.
S11 was measured
The network analyzerinwith
a frequency rangecoaxial
connected from 8 GHz
cablesto 10
wasGHz every 10 to
calibrated 20 seconds.
using a SOLTThen, before kit
calibration
the resonant frequency and the absolute value of
(Anritsu SMA Calibration Kit 3650). The setup is depicted in Figure 6.the reflection coefficient (|S 11|) were determined,

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
external furnace.and the absolute
Therefore, the coatedvalue of the reflection
microwave transducercoefficient
without the(|S SMA 11 |) were determined,
connector was heated the
reflection spectrum around the operating resonant frequency was filtered
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 using a Lorentz fit tokept
remove
noiseatand
600parasitic reflections.
°C for 10 min. Finally, the transducer was cooled down to room temperature over 25 min ( 21
Generally, after each
°C). Furthermore, after adsorption measurement,
storing in ambient the zeolite
air, the microwave had to be
transducer was thermally
purged inregenerated
N2 for severalin an
minutes
external to desorb
furnace. stored the
Therefore, water on themicrowave
coated zeolite (see transducer
also next section).
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 8 of 14
Sensors 2017, 17, 2422 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
Before thebasic concept and the
measurements repeatability
in the of the microwave-based
above-described setup started, adsorption monitoring
initial tests of the to
were performed
zeolites at room temperature. For that purpose, the microwave sensor was connected
verify the basic concept and the repeatability of the microwave-based adsorption monitoring to the VNA andof the
placed in different atmospheric conditions. Apart from ambient air, a rolled-neck glass
zeolites at room temperature. For that purpose, the microwave sensor was connected to the VNA and partly filled
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
with thisinto the gas phase established an NH3 partial pressure in the headspace of the liquid
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:
2. Sensor inside the glass (ammonia-in-air ambience) without zeolite layer.
1. 3. Sensor
Sensor outside
outside thethe glass
glass (i.e.,ininair)
(i.e., air)without
with zeolite layer.
zeolite layer.
2. 4. Sensor
Sensor inside
inside thethe glass
glass (ammonia-in-air ambience)
(ammonia-in-air ambience) with zeolite
without layer. layer.
zeolite
3. Sensor outside
If the sensorthe glass (i.e.,
is placed insideinthe
air)glass
with(inzeolite layer. of the liquid), both gaseous ammonia
the headspace
4. and
Sensor inside the glass (ammonia-in-air ambience)sensors’
water should adsorb in the zeolite and shift the resonant
with zeolite frequency towards lower
layer.
frequencies since the effective permittivity increases. For reference purposes, the experiment was also
If the sensor
repeated is placed
without a zeoliteinside
cover the glass
layer. (in the headspace
In addition, the initial testofcases
the liquid), both gaseous
were simulated with Comsolammonia
and water shouldThe
Multiphysics. adsorb in the
geometry forzeolite
the modelsandwasshift
setthe sensors’
to be the same resonant frequency
as the fabricated ringtowards
resonatorlower
sensor (specifications:
frequencies since the effective substrate thicknessincreases.
permittivity of 0.635 mm, Formean ring radius
reference R = 2.07
purposes, themm, coupling was
experiment gap also
d = 0.15 mm, ring microstrip width W = 0.22 mm, zeolite thickness 0.75
repeated without a zeolite cover layer. In addition, the initial test cases were simulated with Comsol mm). The Al 2 O 3 substrate
properties The
Multiphysics. weregeometry
chosen to beforε’ =the
9.87 and tanwas
models δ = 0.0003,
set towhich
be theis same
close toasthe
thecorresponding
fabricated ring datasheet
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,
permittivity of the there
zeolitewere
and nothe data valuesRoughly
sensitivity. for the they
complex
agreepermittivity
with data from available for the used
[40] for zeolites.
Fe-zeolite
Finally, considering the attenuation of transmission lines in the measurement setup, an attenuation were
and the ammonia-saturated Fe-zeolite. Thus, the complex permittivities of the zeolites
chosen in adB
of 0.55 waywasthat
addedthey fit the
to the measurement
Comsol data best (ε’zeolite,air = 3.35, tan δzeolite,air = 0.07 and
simulation data.
ε’zeolite,NH3 Figure
= 3.97,7 depicts the measured
tan δzeolite,NH3 reflection
= 0.095). Thesespectra
valuesfor canthebefour cases. of
thought Asasexpected, all resonance
a first guess of the actual
peaks differ
permittivity from
of the each other.
zeolite and The
the resonant frequencies
sensitivity. Roughly without a zeolite
they agree layerdata
with in airfrom
and in ammonia-
[40] for zeolites.
in-air are identical (① and ②). This was expected, since the gas phase
Finally, considering the attenuation of transmission lines in the measurement setup, an attenuation above the sensor hardly of
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
are identical ( 1 and ).
2 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 9 of 14
Sensors 2017, 17, 2422 9 of 14
Sensors 2017, 17, 2422 9 of 14

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

Figure 7. 7. Initial
Initial tests:absolute
tests: absolute valueof of thereflection
reflection parameter |S | |ofofthe
themicrowave sensor forfor
Figure
Figure 7. Initial tests: absolutevalue
value ofthe
the reflection parameter
parameter |S |S11
11
11 | of the microwave
microwave sensor
sensor for
different
different atmospheres:solid
atmospheres: solidline
line==measurement;
measurement; circle
circle line
line ==simulation.
simulation. ① Sensor
Sensor outside
outsidethethe
glass
glass
different atmospheres: solid line = measurement; circle line = simulation. ① Sensor outside the glass
1
(i.e.,
(i.e., in in air)
air) withoutzeolite
without zeolitelayer
layer(black dashed); ②
(black dashed); Sensor inside
Sensor inside the
theglass
glass (ammonia
(ammonia inin
airair
ambience)
ambience)
(i.e., in air) without zeolite layer (black dashed); ② Sensor inside the glass (ammonia in air ambience)
2
without zeolite (green); 3③Sensor
layer(green); Sensor outsidethe
the glass (i.e.,
(i.e.,in
inair)
air)with
withzeolite layer (red); ④ Sensor
without
without zeolite
zeolite layer (green);
layer ③ Sensoroutside
outside the glass
glass (i.e., in air) with zeolite
zeolite layer
layer (red);
(red); Sensor
④ 4Sensor
inside
inside the
thethe glass
glass (ammonia
(ammonia in air ambience) with zeolite layer (blue).
inside glass (ammoniaininair airambience)
ambience)with
withzeolite
zeolite layer (blue).
(blue).

The zeolites were regenerated and the experiment was repeated another two times. For each run,
The zeolites wereregenerated
regeneratedand andthetheexperiment
experiment was was repeated another two times. For each run,
theThe zeolites
resonant were
frequency for the unloaded case (after regeneration repeated another
in air), fully two times.
ammonia-loadedFor each run,
case
the resonant
the(ammonia-in-air frequency
resonant frequency for the
for theand unloaded
unloaded case (after regeneration in air), fully ammonia-loaded case
ambience), now alsocase (after regeneration
the partially loaded case in (inair), fullyexposing
air after ammonia-loaded
to ammonia) case
(ammonia-in-air ambience), and now also the partially loaded case (in air after exposing to ammonia)
(ammonia-in-air
were recorded ambience),
and compared. and now alsolatter,
For the the partially
the fullyloaded
loadedcase (in airwere
zeolites afteragain
exposing
exposedto ammonia)
to air.
were recorded and compared. For the latter, the fully loaded zeolites were again exposed to air.
were recorded
Thereby, and compared.
ammonia (definedFor the latter,
as weakly the fully
bonded loaded zeolites
ammonia were again
[44]) desorbs until aexposed to air. Thereby,
new equilibrium is
Thereby, ammonia (defined as weakly bonded ammonia [44]) desorbs until a new equilibrium is
ammonia
established(defined
and onlyas weakly
stronglybonded
bonded ammonia
ammonia [44]) desorbs
remains in until a new equilibrium
the zeolite. The respective is established
resonant
established and only strongly bonded ammonia remains in the zeolite. The respective resonant
and only strongly
frequencies bonded ammonia
are illustrated in Figure remains
8. It shouldin thebe zeolite.
noted that Thethe
respective
experiments resonant frequencies
were carried out atare
frequencies are illustrated in Figure 8. It should be noted that the experiments were carried out at
different in
illustrated days and thus
Figure 8. Itatshould
different be room
noted temperatures
that the as well as humidity.
experiments were Nevertheless,
carried out at in Figure
different days
different days and thus at different room temperatures as well as humidity. Nevertheless, in Figure
and8, the resonant frequencies at the time t for the empty zeolite coincide. Additionally, for all
8, thus
the at different room temperatures as twell as humidity. Nevertheless, in Figure 8, the resonant
0
resonant frequencies at the time 0 for the empty zeolite coincide. Additionally, for all
experimental
frequencies at theruns thet0same
time trends
for the emptycanzeolite
be seen in the frequency shift. The quantitative differences
experimental runs the same trends can be seencoincide. Additionally,
in the frequency for all
shift. The experimental
quantitative runs the
differences
may
same be
trends explained
can be seen by the varying experimental differences in humidity and room temperature.
may be explained by in
thethe frequency
varying shift. Thedifferences
experimental quantitative in differences
humidity and mayroom
be explained
temperature.by the
Furthermore,
varying the unknown
experimental differencesammonia and water
in humidity concentrations
and room temperature. makeFurthermore,
an explanation thedifficult.
unknown
Furthermore, the unknown ammonia and water concentrations make an explanation difficult.
Therefore,
ammonia andfor the subsequent
water concentrations tests,make
the above-described sample chamber
an explanation difficult. Therefore, with
forthe
thegas supply unit
subsequent tests,
Therefore, for the subsequent tests, the above-described sample chamber with the gas supply unit
and
theand defined
above-describedconcentrations was
sample chamber used.
defined concentrations was used.with the gas supply unit and defined concentrations was used.

Figure
Figure Initial
8. 8. tests:
Initial tests:resonant
resonantfrequencies
frequencies of
of the
the microwave sensorfor
microwave sensor forthe
thethree
threeexperimental
experimental runs
runs
Figure 8. Initial tests: resonant frequencies of the microwave sensor for the three experimental runs
with different
with differentloading
loadingconditions
conditionsof
ofthe
thezeolite:
zeolite: empty, partly
partly loaded,
loaded,und
undfully
fullyloaded
loaded (for
(for further
further
with different loading conditions of the zeolite: empty, partly loaded, und fully loaded (for further
details seesee
details text).
text).
details see text).
Sensors 2017, 17,
Sensors 2017, 17, 2422
2422 10
10 of
of 14
14
Sensors 2017, 17, 2422 10 of 14

5.2. Tests
5.2.Tests under
Testsunder Defined
underDefined
DefinedGasGas Exposure
GasExposure
Exposure
5.2.
Since humidity
Since humidity
humidity in in air
in air should
air should affect
should affect the
affect the resonant
the resonant frequency,
resonant frequency, magnitude
frequency,magnitudemagnitudeand and timely
andtimely response
timelyresponse
response
behavior Since were observed during water desorption. A sensor that was stored in air was installed in the
behavior
behavior were
were observed
observed during
during6 waterwater desorption.
desorption. A sensor
A sensor that
that was was stored in air was installed
stored in air was installed in inthe
the
test
testrigrig as
rigas described
asdescribed in Figure
describedininFigure
Figure66and and was
andwas then
wasthen purged
thenpurged
purgedwith with pure
withpure N
pureNN.2.The2. The respective reflection spectra
Therespective
respectivereflection
reflectionspectra
spectra
test
around resonant frequency are shown in Figure 9.9. The blue curve 2 denotes aa spectrum after storage
around
around resonant
resonant frequency
frequency areare shown
shown in Figure
inrecorded
Figure 9. after The
The blue blue curve
curve denotes
denotes spectrum
a spectrum after
after storage
storage in
in
inair.
air. The
The black
black and
and the
the red
red curve
curve were
were recorded after purging
purging with
with NN.22.One
.One
Onecancan clearly
canclearly see
clearlysee that
seethat both
thatboth
both
air.
ffres The black and the red curve were recorded after purging
and |S11| are affected due to the incorporation of water into the zeolite with N 2 during storage in ambient
res and |S11| are affected due to the incorporation of water into the zeolite during storage in ambient
fair.
res and
Since |S 11 |
water are affected
increases due
the to the incorporation
permittivity of the of water
zeolites, into
the the
resonant zeolite during storage
frequency increases in after
ambientthe
air.Since
air. Sincewater
waterincreases
increasesthe thepermittivity
permittivityof ofthe
thezeolites,
zeolites,the theresonant
resonantfrequency
frequencyincreases
increasesafterafterthe
the
desorption
desorption of
of water.
water. The
The results
results also
also show
show that
that after
after 10
10 min
min most
most of
of the
the water
water has
has already
already desorbed,
desorbed,
desorption
even at roomof water. The results
temperature (Figure also show
9b). On that
the after 10
other minitmost
hand, of the
isis clear that water
before haseach
already desorbed,
measurement,
even
even at room
at room temperature
temperature (Figure
(Figure 9b).
9b). On On the
the otherother hand, it
hand, itreproducible clear that
is clear that before before each measurement,
each measurement, the
the
thesensor
sensor has
has to bebepurged for
forseveral minutes to
toobtain results.
sensor has to betopurgedpurged
for several several
minutesminutes
to obtain obtain reproducible
reproducible results. results.

(a)
(a) (b)
(b)

Figure 9. Absolute value

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

Then,
Then,
Then,the the influence
theinfluence
influenceof of defined
ofdefined water
definedwater content
watercontent was
contentwas investigated.
wasinvestigated. Figure
investigated.Figure
Figure10 10 isisaaatime-dependent
10is time-dependent
time-dependent
plot
plot of
of f
fres and
plot of fres and |S|S11|@f
res and |S1111
res for an
|@f
|@fres for for admixture
an of
admixture 1 vol
of 1% and
vol % 2
andvol %
2 of
vol water
% of into
water nitrogen.
into
res an admixture of 1 vol % and 2 vol % of water into nitrogen. One can
One can
nitrogen. clearly
One can
clearly
see that
clearly the
see thatsee device
thethat responds
the responds
device to
device responds water. This
to water.toThis excludes
water. O 2 or
This excludes
excludes CO 2 (both
O2 or CO2O(both from
2 or CO
air)
2 (both
from from being
frombeing
air) from responsible
air) from being
responsible
for
forthe
thesensor
responsiblesensorforeffect in
inFigure
the sensor
effect 9.
9.Again,
effect
Figure in adsorbed
Figure
Again, water
9. Again,
adsorbed can
canbe
adsorbed
water bedesorbed
water canin
desorbed bewater-free
in desorbednitrogen
water-free even
in water-free
nitrogen even
at
at room temperature, but the process requires a longer time. Here, higher temperatures would be
room
nitrogen temperature,
even at room but the
temperature,process
but requires
the a
process longer
requires time.
a Here,
longer higher
time. temperatures
Here, higher would
temperatures be
favorable.
would be favorable.
favorable.

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

In further
Infurther experiments,
furtherexperiments, defined
experiments,defined
definedNH NH
NH 3 concentrations were admixed. The sensor was exposed to
3 concentrations wereadmixed.
admixed. The sensor was exposed
In 3 concentrations were The sensor was exposed to to
a
aastepwise-changing
stepwise-changing NH
NH 3 concentration from 0 ppm to 500 ppm (t1) and from 500 ppm to 1000 ppm
3 concentration from0 0ppmppmtoto500
500ppm
ppm(t(t
stepwise-changing NH 3 concentration from 1 )1)and
andfromfrom500
500ppmppmtoto10001000ppm
ppm
NH
NH333 (t 2) without regenerating the zeolite in between. Again, the obtained results in Figure 11 show
NH (t(t22)) without
without regenerating
regenerating the the zeolite
zeolite in
in between.
between. Again,
Again, the
the obtained
obtained results
results in
inFigure
Figure11 11show
show
different resonant
differentresonant frequency
resonantfrequency shifts
frequencyshifts during
shiftsduring
duringthethe loading
theloading with
loadingwith 500
with500 ppm
500ppm
ppmand and 1000
and1000 ppm
1000ppm
ppmNH NH
NH33.3..In
In the
Inthe first
thefirst
first
different
measurement
measurementstep step between
stepbetween t 1 and t2, when loading with 500 ppm NH3, it can be clearly seen that the
betweent1t1andandt2t2,,when
when loading
loadingwith
with500
500ppm
ppmNH NH33,, itit can
canbebeclearly
clearlyseen
seenthatthatthe
the
measurement
sensor
sensorsignalsignalisisaccumulating
accumulatinguntil untilsaturation
saturationwithwithammonia
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
ammonia concentration to to 1000
1000 ppm
ppm alsoalso leads to aa distinguishable
leads to distinguishable but but smaller
smaller resonant
resonant frequency
frequency
shift. The differences in the sensor signal between the 500 ppm saturation
shift. The differences in the sensor signal between the 500 ppm saturation and the 1000 ppm saturation and the 1000 ppm
saturation
is due to the is due
higherto the higher
partial partial pressure
pressure of NH3 that of NH 3 thatacauses
causes a shift towards
shift towards the adsorbed
the adsorbed phase phase
in the
in the adsorption
adsorption isotherm. isotherm.
The feedThe with
feed 1000
with ppm1000 ppm
ammoniaammonia is finally
is finally stopped
stopped andgas
and the the sensor
gas sensor
test
test chamber is purged with pure N
chamber is purged with pure N2 (t3 ). Consequently, the resonant frequency shift decreases, but notbut
2 (t 3). Consequently, the resonant frequency shift decreases, to
not initial
the to the value
initialas value as probably
probably expected expected fromdesorption.
from water water desorption. It is assumed
It is assumed that only that only so-called
so-called weakly
weakly NH
bonded bonded NH3 can desorb at room temperature. Strongly bonded ammonia remains on the
3 can desorb at room temperature. Strongly bonded ammonia remains on the zeolite even
in a pure nitrogenpure
zeolite even in a nitrogen
atmosphere at atmosphere at this low
this low temperature. temperature.
These results agreeThese
wellresults
with dataagree well with
obtained for
>200 ◦
data obtained
C for an for >200 °C
H-ZSM5 for anobtained
zeolite H-ZSM5inzeolite obtained
an electrical in an
in situ electrical
cavity in situ
resonator cavity
[40], resonator
if one [40],
extrapolates
if one
their dataextrapolates their dataHowever,
to room temperature. to roomwith temperature. However, applications,
respect to gas-sensing with respectit is toclear
gas-sensing
that one
applications,
needs a higher it is clear that one
temperature needs athe
at which higher temperature
desorption is fast at whichto
enough the desorption
guarantee is fast enough
a reversible sensorto
guaranteeTherefore,
behavior. a reversible it issensor behavior.
mandatory Therefore,
to establish it is mandatory
a heater layer on theto establish
sensor a heater
substrate. Thislayer
will on
be the
the
sensor
next stepsubstrate.
during the Thissensor
will be the next step during the sensor development.
development.

Figure 11. Time dependency

dependencyof ofthe
theresonant
resonantfrequency
frequencyofofthethemicrowave
microwave sensor
sensor according
according to to Figure
Figure 4c
4c when
when 500500
ppmppm
(t1 )(tor
1) or 1000
1000 ppmppmNH NH (t
3 23 (t
) are
2 ) are added
added to to
N N
2 . .
At
2 At
t 3 t
, 3 ,
NH NH
3 was
was
3 turned
turned off.
off.

6. Conclusions
6. Conclusions and
and Outlook
Outlook
This work investigated
This work investigated the design parameters
the design parameters of of aa microstrip
microstrip ringring resonator
resonator with
with anan Fe-zeolite
Fe-zeolite
as aa gas-sensitive
as gas-sensitive layer.
layer. Based
Based on on FEM simulation results
FEM simulation and commonly
results and commonly used used microstrip
microstrip design
design
equations, aa planar
equations, planar microstrip
microstripring
ringresonator
resonatorstructure
structureon onalumina
aluminawas wasdeveloped
developedand andcovered
coveredwithwitha
a Fe-zeolite film. Finally, the device was successfully operated at around 8.5
Fe-zeolite film. Finally, the device was successfully operated at around 8.5 GHz as a humidity sensorGHz as a humidity
sensor
and forand for ammonia-loading
ammonia-loading detection detection
at room at room temperature.
temperature. The amount Theof amount
stored of stored
water andwater and
ammonia
ammonia in the zeolite is mirrored by both the absolute value of the reflection coefficient
in the zeolite is mirrored by both the absolute value of the reflection coefficient at resonance and by at resonance
andresonant
the by the resonant
frequency frequency
itself. Dueitself. Due
to the to the
better better resolution,
resolution, we conclude we that
conclude that thefrequency
the resonant resonant
frequency
is a suitableismeasure
a suitable measure for
for ammonia and ammonia and water
water detection with thisdetection
planar with thisFurthermore,
device. planar device. the
concentration-dependent frequency shifts during loading illustrate the accumulating characteristicsthe
Furthermore, the concentration-dependent frequency shifts during loading illustrate of
accumulating
the sensor signalcharacteristics
and make itofconceivable
the sensor tosignal and make
measure smallitamounts
conceivable to concentrations
of gas measure small by amounts
using
of gas
the concentrations
device operating in bytheusing the device
so-called operating in[45].
dosimeter-mode the so-called dosimeter-mode [45].
In
In future work, an integrated heating element willbebeadded
future work, an integrated heating element will addedtotothe thedevice.
device.Then
Thenit may
it may work
workas
a fast
as andand
a fast reversible
reversiblegasgas
sensor.
sensor.In addition,
In addition,it may
it mayhelp to determine
help to determine adsorption
adsorptionisotherms. In
isotherms.
combination with in situ DRIFT spectroscopy, the effect of distinct adsorbed species
In combination with in situ DRIFT spectroscopy, the effect of distinct adsorbed species on the dielectric on the dielectric
behavior of
behavior of the
the gas
gas sensitive
sensitive material
material cancan also
also be
be studied
studied inin aa wide
wide temperature
temperature regime.
regime.

Acknowledgments: The authors would like to thank M. Dietrich and D. Rauch for the constructive discussions
and
and some
some technical
technical assistance. The publication
assistance. The publication of
of this
this paper
paper was
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,
Sensors 2017, 17,
17, 2422
2422 12 of
12 of 14
14
Conflicts of Interest: The authors declare no conflict of interest.

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

Figure A1. Microstrip waveguide: (a) cross section with ground plane and strip conductor; (b) electric
Figure A1.
Figure A1. Microstrip
Microstrip waveguide:
waveguide: (a)(a) cross
cross section
section with
with ground plane and strip conductor; (b) electric
(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
Simplified setup for
for the
the initial
initial measurements.
initial measurements
measurements

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

References
References
1.
1. Comini, E.;
Comini,
Comini, E.; Faglia,
Faglia, G.; Sberveglieri,
Faglia, G.; Sberveglieri,
Sberveglieri, G. G. Solid
G. Solid State Gas
Solid State Gas Sensing;
Sensing; Springer
Springer International
Springer International Publishing
International Publishing AG:
Publishing AG:AG:
Heidelberg, Germany,
Heidelberg, Germany,
Germany, 2009. 2009;
2009; doi:10.1007/978-0-387-09665-0.
doi:10.1007/978-0-387-09665-0.
2.
2. Bănică, F.-G.
Bănică, F.-G. Chemical
Chemical Sensors
Sensors and
and Biosensors:
Biosensors: Fundamentals
Biosensors: Fundamentals
Fundamentals and and Applications;
and Applications; John
Applications; John Wiley
John Wiley
Wiley & & Sons,
& Sons, Ltd.:
Sons, Ltd.:
Ltd.:
Chichester, UK, 2012; pp.
Chichester, UK, 2012; pp. 246–257.246–257, doi:10.1002/9781118354162.
246–257, [CrossRef]
doi:10.1002/9781118354162.
3.
3. Yamazoe, N.;
Yamazoe, N.; Shimanoe,
Shimanoe, K. K. Fundamentals
Fundamentals of of Semiconductor
Semiconductor Gas Gas Sensors.
Sensors. InIn Semiconductor
In Semiconductor
Semiconductor GasGas Sensors;
Gas Sensors;
Sensors;
Jaaniso, R.,
Jaaniso, R., Tan,
Tan, O.K.,
O.K., Eds.;
O.K., Eds.; Woodhead
Eds.; WoodheadPublishing
Woodhead PublishingLtd.:
Publishing Ltd.:Cambridge,
Ltd.: Cambridge,UK,
Cambridge, UK,2013;
UK, 2013;pp.
2013; pp.3–34.
pp. 3–34.
3–34.
4.
4. Barsan, N.;
Barsan, N.; Koziej,
Koziej, D.;
D.; Weimar,
Weimar,
Weimar, U. U. Metal
U. Metal Oxide-Based
Oxide-Based Gas
Oxide-Based Gas Sensor
Gas Sensor Research:
Sensor Research: How
Research: How to? to? Sens.
Sens. Actuators
Actuators
Actuators BB Chem.
B Chem.
Chem.
2007, 121, 18–35,
2007, 121, 18–35, doi:10.1016/j.snb.2006.09.047.
18–35. doi:10.1016/j.snb.2006.09.047.
[CrossRef]
5.
5. Zarifi, M.H.;
Zarifi, M.H.; Shariaty,
Shariaty, P.;P.; Hashisho,
P.; Hashisho, Z.;
Hashisho, Z.; Daneshmand,
Z.; Daneshmand, M.
Daneshmand, M. AA Non-Contact
Non-Contact Microwave
Microwave SensorSensor for
for Monitoring
Monitoring
the Interaction
the Interaction of of Zeolite
Zeolite 13X13X with
with COCO and CH
CO22 and CH44 in in Gaseous
Gaseous Streams.
Streams. Sens.
Sens. Actuators
Actuators
Actuators B B Chem.
B Chem. 2017,
Chem. 2017, 238,
2017, 238,
238,
2 4
1240–1247, doi:10.1016/j.snb.2016.09.047.
1240–1247, doi:10.1016/j.snb.2016.09.047.
1240–1247. [CrossRef]
6.
6. Zarifi, M.H.;
Zarifi, M.H.; Sohrabi,
Sohrabi,
Sohrabi, A.;A.; Shaibani,
A.; Shaibani, P.M.;
Shaibani, P.M.; Daneshmand,
P.M.; Daneshmand,
Daneshmand, M.; M.; Thundat,
M.; Thundat,
Thundat, T.T. Detection
T. Detection
Detection of of Volatile Organic
of Volatile Organic
Compounds Using Microwave Sensors. IEEE Sens.
Compounds Using Microwave Sensors. IEEE Sens. J. 2015, 15, 248–254, J. 2015, 15, 248–254, doi:10.1109/JSEN.2014.2345477.
248–254. doi:10.1109/JSEN.2014.2345477.
[CrossRef]
7.
7. Rossignol, J.;
Rossignol, J.; Barochi,
Barochi, G.;
G.; de
de Fonseca,
Fonseca,
Fonseca, B.;B.; Brunet,
B.; Brunet, J.;
Brunet, J.; Bouvet,
J.; Bouvet, M.;
Bouvet, M.; Pauly,
M.; Pauly, A.;
Pauly, A.; Markey,
A.; Markey,
Markey, L. L. Microwave-based
Microwave-based gas gas
sensor with
sensor with phthalocyanine
withphthalocyanine
phthalocyanine film
film at
at room
film room
at room temperature.
temperature.
temperature. Sens. Actuators
Sens. Actuators B Chem.
B Chem. B2013,
Sens. Actuators 2013,
189,2013,
Chem. 189,
213–216. 213–216,
189,[CrossRef]
213–216,
doi:10.1016/j.snb.2013.03.092.
doi:10.1016/j.snb.2013.03.092.
8.
8. Bailly, G.;
Bailly, 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
Interdigital Capacitor:
Interdigital Capacitor: Detection
Detection of of NH with TiO
NH33 with TiO22 nanoparticles.
nanoparticles. Sens.
Sens. Actuators
Actuators B B Chem.
Chem. 2016,
2016, 236,
236, 554–
554–
564, doi:10.1016/j.snb.2016.06.048.
564, doi:10.1016/j.snb.2016.06.048.
9.
9. Bailly, G.;
Bailly, 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:
Shape Effect on Ammonia Detection Using Pseudocubic, Rhombohedral, and Spindlelike Particles. ACS
Sensors 2017, 17, 2422 13 of 14

8. 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]
9. 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]
10. 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.
11. 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]
12. 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]
13. Sarabandi, K.; Li, E.S. Microstrip Ring Resonator for Soil Moisture Measurements. IEEE Trans. Geosci.
Remote Sens. 1997, 35, 1223–1231. [CrossRef]
14. 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]
15. 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]
16. Zarifi, M.H.; Thundat, T.; Daneshmand, M. High resolution microwave microstrip resonator for sensing
applications. Sens. Actuators A Phys. 2015, 233, 224–230. [CrossRef]
17. 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]
18. 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]
19. 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]
20. Xu, X.; Wang, J.; Long, Y. Zeolite-based Materials for Gas Sensors. Sensors 2006, 6, 1751–1764. [CrossRef]
21. 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]
22. 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]
23. 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]
24. 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]
25. 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.
26. 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.
27. 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]
28. 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 14 of 14

29. 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]
30. 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]
31. Pozar, D.M. Microwave Engineering, 4th ed.; John Wiley & Sons, Ltd.: Hoboken, NJ, USA, 2012.
32. 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]
33. 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]
34. 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]
35. 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]
36. 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]
37. Moos, R. Microwave-Based Catalyst State Diagnosis—State of the Art and Future Perspectives. SAE Int.
J. Engines 2015, 8, 1240–1245. [CrossRef]
38. Hoffmann, R.K. Integrierte Mikrowellenschaltungen; Springer: Berlin/Heidelberg, Germany, 1983.
39. 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]
40. 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]
41. 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]
42. 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]
43. 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]
44. 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]
45. 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/).