Instrumentation & Measurement Magazine 25-3 - 45

For the sake of completeness, it should be mentioned that
the integration of microwave propagation structures with microfluidic
channels enables experimental characterization of
microliter and even nanoliter volumes of liquids [13].
Sensing Materials/Gas Detection
Among different materials [14], metal oxides constitute the
conventional sensing layers for microwave gas detectors. Recently,
titanium dioxide (TiO2
oxygen detection [6]. A thick layer of TiO2
) was used as sensing layer for
is deposited by
screen printing upon an interdigital capacitor that acts as a
microwave transducer. The resonant frequencies are used as
tracker for oxygen detection. Similarly, iron oxide (Fe2
O3
) is
employed for water vapor detection by depositing the metal
oxide upon a microwave transducer [10].
Zeolites and nanostructured materials are also preferred
for microwave gas sensors because of their unique properties.
Their special structure allows the gas molecules to be absorbed
in their volume. In general, the sensitivity of a microwave gas
detector is proportional to the sensing material surface exposed
to the gas. By using a porous material like zeolite or a
nanostructured material instead of a bulk material, the exposed
surface is increased considerably, and consequently, the
sensitivity is improved by six/eight orders of magnitude [15].
Moreover, the sensor response becomes much faster. Carbon
nanotubes (CNTs) have been successfully used for ammonia
detection. In [15], a CNT-based sensing layer is deposited upon
a microstrip disk resonator, and the device exhibited a high
sensitivity (4 kHz/ppm). Zeolite layers have been also used
for toluene detection. De Fonseca et al. [16] employ zeolite as
sensitive material by spin coating deposition upon coplanar
microwave structures. The device is able to detect toluene at
ppm concentrations working at ambient temperature.
The best place to deposit the sensing material on the propagative
structure is where it has the greatest effect in terms of
sensitivity when its dielectric properties change as a consequence
of the gas adsorption. This depends on the selected
resonant mode and in particular to its electromagnetic field
configuration. The sensing layer can cover the whole resonator
or only a part of it, and it can be deposited on coplanar lines or
in the coupling gap between the microwave resonator and the
feedline [10]. The main goal consists of maximizing the sensor
response when the target gas is detected.
Sensor Response and Case Studies
The sensing performance of a microwave gas sensor is evaluated
using a vector network analyzer (VNA). This instrument
describes the behavior of a device under test (DUT) in terms
of scattering (S-) parameters, allowing determination of key
quantities like the return loss and the input impedance. The
alteration of the sensing performance can be seen as variations
of the resonant frequency in the microwave spectrum or
into amplitude variations (A) at a fixed frequency or into variation
of the quality factor (Q). The estimation of the triplet fr
,
A, and Q from a discrete frequency spectrum is a not trivial
task, especially if the data are affected by noise. A simple linear
May 2022
interpolation of the acquired points could result in a low accuracy
estimation of these quantities. In order to overcome this
issue, a Lorentzian fitting of the frequency spectrum is usually
preferred [10], since it is possible to estimate fr
, A, and Q with
relatively low uncertainty from the fitted function. It is also
important to evaluate the dynamic performance of the sensor,
including the response time and recovery time. In this case,
the evolution in time of these quantities is recorded and their
change is evaluated.
As a first case study, a concentric rings microstrip (CRM)
resonator is considered [10] (Fig. 2a). This device is proposed
as an alternative to the classic ring resonator to reach a higher
quality factor by more than +40% [10].
A sensing material based on iron oxide (Fe2
O3) was deposited
on the coupling gap between the resonator and the
feedline. The device was tested at room temperature as a water
vapor detector. Since the considered microwave resonator has
only one port, the reflection coefficient was used for humidity
detection. The variation of S11
at different relative humidity
,
(RH) concentrations is reported in Fig. 2b. The parameters fr
A, and Q, calculated by the Lorentzian fitting procedure, are
affected by the water vapor adsorbed on the surface of the
sensing material; however, in this paper only frequency shift is
considered for tracking the RH. It is worth noting that, in this
case, the Lorentzian fitting is applied only to the S11
magnitude.
A more accurate result can be obtained by using a complex Lorentzian
function to fit both the magnitude and phase of the
acquired spectrum.
The CRM resonator is tested in a controlled atmosphere in
which the RH is changed from dry air up to 83% RH (Fig. 2c).
The two resonant frequencies (i.e., f1
and f2
= 3.7 GHz with Q1
= 4500
= 5.4 GHz with Q2 = 5000) are found to be strongly related
with the water vapor content so that it was possible to
employ the CRM resonator as a humidity detector. The exponential
function was used to fit the acquired points with a R2
of 0.994 and 0.956 for Dip1 and Dip2, respectively. Such values
are in line with those reported in the literature for the other typologies
of gas sensors.
The CRM sensor and, in general, sensors using nanomaterials,
tend to saturate toward a peak as the analyte molecules
complete their reaction with the sensor surface. Therefore,
the desorption to baseline becomes weaker with increasing
humidity levels, i.e., all the adsorbed gas molecules are not released
when dry air is injected onto the sensor surface. That is
why smaller variations in resonant frequency are present at
the higher humidity levels. Hence, the response appears piecewise
linear with two slopes, a steep one for 40% or less RH and
a much smaller slope for RH > 40%.
Further information on this research activity can be found
in [10], whereas a detailed description of the used sensing material
is given in [17].
The second case study considered here is a two-port IDC
covered by a thick layer of TiO2 as sensing material and employed
in oxygen (O2) detection (Fig. 3a) [18]. The device is
exposed to different O2 concentrations in the range from 0% to
100% and the sensor response was studied by measuring the
IEEE Instrumentation & Measurement Magazine
45
Instrumentation & Measurement Magazine 25-3

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-3
