sensors

Article
Microfluidic EBG Sensor Based on Phase-Shift
Method Realized Using 3D Printing Technology
Vasa Radonić *, Slobodan Birgermajer and Goran Kitić
BioSense Institute—Research Institute for Information Technologies in Biosystems, Dr Zorana Đinđića 1a,
21000 Novi Sad, Serbia; [email protected] (S.B.); [email protected] (G.K.)
* Correspondence: [email protected]; Tel.: +381-21-485-21-38

Academic Editors: Jae-Won Choi and Erik D. Engeberg

Received: 14 February 2017; Accepted: 17 April 2017; Published: 18 April 2017

Abstract: In this article, we propose a novel microfluidic microstrip electromagnetic band gap (EBG)
sensor realized using cost-effective 3D printing technology. Microstrip sensor allows monitoring
of the fluid properties flowing in the microchannel embedded between the microstrip line and
ground plane. The sensor’s operating principle is based on the phase-shift method, which allows
the characterization at a single operating frequency of 6 GHz. The defected electromagnetic band
gap (EBG) structure is realized as a pattern in the microstrip ground plane to improve sensor
sensitivity. The designed microfluidic channel is fabricated using a fused deposition modelling
(FDM) 3D printing process without additional supporting layers, while the conductive layers are
realized using sticky aluminium tape. The measurement results show that the change of permittivity
of the fluid in the microfluidic channel from 1 to 80 results in the phase-shift difference of almost
90◦ . The potential application is demonstrated through the implementation of a proposed sensor
for the detection of toluene concentration in toluene–methanol mixture where various concentrations
of toluene were analysed.

Keywords: microstrip sensor; electromagnetic band gap (EBG); microfluidics; 3D printing;

fused deposition modelling (FDM); phase-shift method

1. Introduction
Microfluidics is a technology of manipulating the small quantity of fluids in the range
of microliters to picoliters in one or a network of microchannels. Since microfluidic technology allows
for operation and control of fluids on a sub-microscale, it found the applications in various scientific
and engineering disciplines such as inkjet printing, chemistry, environment, biomedicine, etc. [1–5].
Nowadays, advanced microfluidic biochips integrate into a single chip a number of operations such
as sample pre-treatment and preparation, cell separation, and transport, mixing and/or separation
of fluids together with micromechanical, optical, and electronic components for sensing and detection.
In order to integrate advance functions into a singular chip, modern microfluidics combine
multiple technologies including microelectromechanical systems technology, injection moulding,
photolithography and X-ray lithography, laser ablation and micromachining, etc. [1,4,5]. However,
all stated microfabrication techniques are relatively complex, time-consuming processes that typically
require additional manual manufacturing procedures.
Nowadays, 3D printing technology attracts significant attention due to its low-cost, simple
fabrication process that can be realized in a single run, good system compatibility, and presence
of a number of different materials with good optical, biocompatible, chemical or mechanical
properties [6–8]. The 3D printing offers the opportunity to fabricate the whole microfluidic device
in a single run without the need for additional assembly processes. A wide range of biomaterials,
such as living cells and growth factors, could also be directly printed using 3D printing technology [8].

Sensors 2017, 17, 892; doi:10.3390/s17040892 www.mdpi.com/journal/sensors

Sensors 2017, 17, 892 2 of 14

The 3D printing technology is based on different manufacturing methods, such as stereolithography,

multi-jet modelling, electron beam melting, bioprint, and fused deposition modelling (FDM). The most
commercially available 3D printers operate according to the FDM method, which has relatively low
accuracy and speed in comparison to other methods. FDM 3D printers build structure layer-by-layer
from the bottom up by extruding thermoplastic filament through a heated nozzle and deposit it in fine
threads along the extrusion path. Recently, FDM technique has been employed to produce LEDs,
sensors, antennas, and electrodes within biological tissue [6,7].
Fabrication of microfluidic channels by FDM is still a challenge because of several limitations:
extruded filaments cannot be arbitrarily joined at intersections, the lack of structural integrity between
the layers results in weak seals, and the size of the extruded filament is larger than a typical channel
used in microfluidics. So far, the utilisation of FDM 3D printing technology for applications in different
microfluidic devices has been reported in a number of publications in literature [6–15]. A reactor
with fluidic 3 mm tube was fabricated using acrylonitrile butadiene styrene (ABS) polymer in [9],
but this device suffers from leakage and low operation pressure. Microfluidic channel of 800 µm
was used in the realization of organic and inorganic synthesis reactionware for electrochemical and
spectroscopic analysis [10]. Furthermore, fluidic devices with the same size of the microchannel were
used for nanoparticle preparation and electrochemical sensing [11], and detection of cancer protein
biomarkers based on supercapacitor-powered immunoarray [12]. However, it is still challenging
to fabricate leak-proof microfluidic channels narrower than 800 µm with the help of commercially
available filaments using FDM 3D printing technology. In [13], a custom FDM printer was designed
for enhanced resolution of the microchannel, where 350 µm wide polycaprolactone microchannels
have been applied in the realization of 3D nervous-system-on-a-chip.
Recent studies on microfluidics have been expanded to fluid detection at radio frequency (RF)
and microwave frequencies [14–20]. RF and microwave microfluidics use fluids as a substitutive
dielectric material for microwave antennas [14,15], transmission lines [16], or resonators [17].
In that manner, the characteristic parameters of the fluids were determined based on measured
impedance [15,16], resonant frequency [17–19], insertion loss, or the phase of the signal that propagates
along the transmission line [20]. For the sensing applications, the most promising method is the method
based on the phase measurement of the propagating signal [20,21], since it has a relatively fast response,
allows for characterization at single frequency, and it is the least sensitive to insertion loss.
In this paper, we propose a novel microfluidic microstrip sensor realized using cost-effective
3D printing technology. The 350 µm wide microfluidic channel embedded in the microstrip substrate
has been fabricated by conventional FDM 3D printing technology using polylactic acid (PLA)
filament without any supporting layer or soluble material. Leak-proof structure has been ensured
through careful optimization of the 3D model, infill factor and high infill/perimeter overlap settings.
The proposed sensor is designed to allow monitoring of the dielectric properties of the fluid that flows
in the microfluidic channel embedded between the microstrip line and ground plane realized using
defected electromagnetic band gap (EBG) structure. A concept to improve microstrip sensor sensitivity
based on defected EBG structure was proposed in [21], where it is demonstrated that the sensor
sensitivity can be increased by reducing the wave group velocity using periodic patterns in the ground
plane, which exhibits EBG effect [22]. The characteristics of different fluids in the microfluidic channel
are analysed by the phase-shift measurements of the transmitted signal.
The potential application is demonstrated through the realization of the sensor for detection
of toluene concentration in toluene–methanol mixture. Toluene is an aromatic hydrocarbon solvent
and has numerous commercial and industrial applications, such as solvent in paints, thinners and glues.
It is also used for printing and leather tanning processes, for disproportionation to a mixture of benzene,
methanol and xylene, and in the production of a number of synthetic drugs. Toluene, methanol and
their mixtures appear in a liquid state at room temperature. However, not only their liquids but also
their vapour concentrations in the air can become extremely high and can have negative effects on work
environments or can easily burn or explode [23,24]. Hence, the existence of a low-cost, high-sensitive
Sensors 2017, 17, 892 3 of 14

sensor that is able to detect the exact concentrations of toluene or methanol in their mixtures is essential
in their operations, storage or transportation. The advantage of the proposed design has been
Sensors 2017, 17, 892 3 of 13
demonstrated through comparison between the fabricated sensor and other recently published
microfluidic sensors which
in their mixtures operate
is essential according
in their to thestorage
operations, phase-shift method. The advantage of the
or transportation.
proposed design has been demonstrated through comparison between the fabricated sensor and
2. Sensor Design
other recently published microfluidic sensors which operate according to the phase-shift method.
The 3D layout of the proposed microfluidic microstrip sensor with defected EBG etched
in the2.ground
Sensor Design
plane is shown in Figure 1a. The designed sensor consists of three main parts:
a microstrip linelayout
The 3D in theoftop thelayer,
proposed microfluidicchannel
a microfluidic microstripin sensor with defected
the middle, EBG etched
and ground planein the
patterned
with defected EBG structure in the bottom layer. The top layer, shown in Figure 1b, consists of 50 Ω
ground plane is shown in Figure 1a. The designed sensor consists of three main parts: a microstrip
line in line
microstrip the top layer,
with a microfluidic
length Lstrip andchannel
width w, in the middle, and
respectively, andground plane patterned
two tapers. To avoid with
thedefected
short circuit
EBG structure in the bottom layer. The top layer, shown in Figure 1b, consists of 50 Ω microstrip line
between end-launched sub-miniature A (SMA Southwest Microwave 292-04A-5) connectors’ ground
with length Lstrip and width w, respectively, and two tapers. To avoid the short circuit between end-
and the microstrip line, two tapers were designed according to the manufacturer’s recommendation.
launched sub-miniature A (SMA Southwest Microwave 292-04A-5) connectors’ ground and the
The microfluidic channel bent in the shape of meander is embedded in the microstrip substrate
microstrip line, two tapers were designed according to the manufacturer’s recommendation. The
above the defectchannel
microfluidic in the EBG,
bent ini.e.,
thethe most
shape sensitiveislocation
of meander embedded ininthe
thedesign, Figure
microstrip above the a, b
1c. Parameters
substrate
and cdefect
denote the dimensions of meandered microchannel, while w and
in the EBG, i.e., the most sensitive location in the design, Figure c1c. Parameters hc are the width and height
a, b and c denote
of thethe
channel, respectively.
dimensions of meanderedThe total thicknesswhile
microchannel, of thewcsubstrate
and hc arewith embedded
the width microfluidic
and height channel is
of the channel,
respectively.
denoted as h. TheThe total thickness
bottom layer of theof the substrate
sensor, shown with
in embedded
Figure 1d, microfluidic
represents the channel
ground is denoted as
plane realized
usingh.defected
The bottom EBGlayer of the sensor,
structure, shownstructure
periodical in Figure that
1d, represents
consists of the ground
etched planewith
holes realized using dEBG
diameter
placeddefected EBG structure,
at distance pEBG . Theperiodical
defect instructure that consists
EBG is realized under of etched holes with diameter
the microchannel by removaldEBG placed
of the one
at distance pEBG. The defect in EBG is realized under the microchannel by removal of the one periodic
periodic element. Introduction of the defected EBG structure improves sensitivity of the sensor
element. Introduction of the defected EBG structure improves sensitivity of the sensor in comparison
in comparison to the conventional microstrip line. In addition to EBG, two holes that serve as inlet
to the conventional microstrip line. In addition to EBG, two holes that serve as inlet and outlet of the
and outlet
channel ofwere
the channel
designedwere in thedesigned in the
bottom layer forbottom
assembly layer for assembly
of microfluidic of microfluidic
equipment to inject equipment
fluids
to inject fluids into the channel,
into the channel, Figure 1d. Figure 1d.

Figure 1. The layout of the proposed microfluidic electromagnetic band gap (EBG) sensor: (a) 3D
Figure 1. The layout of the proposed microfluidic electromagnetic band gap (EBG) sensor: (a) 3D
view; (b) top layer; (c) substrate with embedded microfluidic channel; (d) bottom layer with defected
view; (b) top layer; (c) substrate with embedded microfluidic channel; (d) bottom layer with defected
EBG structure.
EBG structure.
Figure 2 shows the electrical filed distribution in the orthogonal cross section of the proposed
sensor and comparison of the intensity of the electrical field along the microstrip line. The strongest
electric field exists in the substrate between the microstrip line and the ground plane, i.e., the zone
Sensors 2017, 17, 892 4 of 14

Figure 2 shows the electrical filed distribution in the orthogonal cross section of the proposed
sensor and comparison of the intensity of the electrical field along the microstrip line. The strongest
electric field
Sensors 2017,exists
17, 892 in the substrate between the microstrip line and the ground plane, i.e.,4 of the
13 zone
where the microfluidic channel is located. Moreover, in comparison with a conventional microstrip
line, where the microfluidic
the intensity channel field
of the electric is located. Moreover,
is stronger in in
thecomparison
vicinity ofwith a conventional
a defect microstrip
in the EBG. Therefore,
the changes of the dielectric constant of the liquid that flows in the channel will have the highestthe
line, the intensity of the electric field is stronger in the vicinity of a defect in the EBG. Therefore, impact
to thechanges
sensorofresponse.
the dielectric constant
In order of the liquid
to further that flows
increase in the channel
the sensitivity willsensor,
of the have the highest impact
microfluidic channel
to the sensor response. In order to further increase the sensitivity of the sensor, microfluidic channel
is bent in the shape of the meander without changing the cross-section dimensions of the channel.
is bent in the shape of the meander without changing the cross-section dimensions of the channel. In
In that manner, the effective area of the fluid exposed to the strongest electrical field increases compared
that manner, the effective area of the fluid exposed to the strongest electrical field increases compared
to thetotopologies
the topologies thatthat
useuse
straight
straightmicrofluidic
microfluidic channel.
channel.

Figure
Figure 2. 2.Electric
Electric field
field distribution
distributionof of
thethe
proposed microfluidic
proposed EBG sensor:
microfluidic EBG (a) orthogonal
sensor: cross-
(a) orthogonal
section view; (b) comparison of the intensity of the electrical field along the microstrip
cross-section view; (b) comparison of the intensity of the electrical field along the microstrip line for line
conventional microstrip line and microstrip line with
for conventional microstrip line and microstrip line with EBG. EBG.

3. Sensor Operating Principle: Phase-Shift Method

3. Sensor Operating Principle: Phase-Shift Method
Phase-shift method is based on a measurement of phase delay of the sinusoidal signal that
Phase-shiftalong
propagates method is based on
the transmission line.aThe
measurement
phase shift of of
thephase delay
signal is of thebysinusoidal
determined velocity andsignal
that propagates
frequency of along theastransmission
the signal line. The
well as the physical phase of
properties shift
the of the signalline:
transmission is determined by velocity
and frequency of the signal as well as the physical properties of the transmission line:
∆ = , (1)
TL ωL
where ω denotes angular velocity, vp is the∆ϕ phase
= velocity , of the signal, and LTL is the transmission (1)
v p
line effective length. On the other hand, the phase velocity of the signal is dependent on the properties
of the surrounding medium of the transmission line. In general, the phase velocity of the signal can
where ω denotes angular velocity, vp is the phase velocity of the signal, and LTL is the transmission
be determined as:
line effective length. On the other hand, the phase velocity of the signal is dependent on the properties
√
of the surrounding medium of the transmission = ,
line. In general, the phase velocity of the signal can
√ (2)
be determined as: √
2 1
where μ is magnetic permeability,vεp is √ r permittivity
= dielectric , σ is electric conductivity of the (2)
and
µε q
σ2
1 + 1 + ω 2 ε2
medium that surrounds the transmission line, [25]. If the operating frequency is high enough, the
influence of conductivity can be neglected and expression for phase velocity can be reduced to:
where µ is magnetic permeability, ε is dielectric permittivity and σ is electric conductivity of the medium
= , (3)
that surrounds the transmission line, [25]. If the operating
√ frequency is high enough, the influence
of conductivity can bedepends
which exclusively neglected
on and expression
permittivity and for phase velocity
permeability can be reduced
of surrounding to:In this way,
medium.
cross-sensitivity with respect to conductivity can be avoided. This is particularly interesting if only
the change in permittivity needs to be monitored, 1
v = as √in the
p , case of detecting pollutants in water [26]. (3)
µε
Sensors 2017, 17, 892 5 of 14

which exclusively depends on permittivity and permeability of surrounding medium. In this way,
cross-sensitivity with respect to conductivity can be avoided. This is particularly interesting if only
Sensors 2017, 17, 892 5 of 13
the change in permittivity needs to be monitored, as in the case of detecting pollutants in water [26].
The Theproposed
proposedsensor
sensor is based on
is based ona atransmission
transmissionlineline in form
in the the form of microstrip
of microstrip architecture
architecture with
with
degeneration in the ground plane and inhomogeneous surrounding medium, Figure Figure
degeneration in the ground plane and inhomogeneous surrounding medium, 3. The 3.
Thesurrounding
surrounding medium
medium of microstrip
of the the microstrip line consists
line consists of from
of the air the air
topfrom topmicrofluidic
and the and the microfluidic
channel
channel embedded in 3D printed
embedded in 3D printed substrate. substrate.

Figure 3. Cross section of the proposed sensor.

Figure 3. Cross section of the proposed sensor.

In order to apply the above-stated equations, the concept of effective permittivity can be
In order In
introduced. tothe
apply thecase
simple above-stated
of the microstrip equations,
line, thethe concept
effective of effective
permittivity can bepermittivity
approximated can
be as:
introduced. In the simple case of the microstrip line, the effective permittivity can
be approximated as:
ε a +=ε s f ε+a − ε s f 1, (4)
εe f f = + q , (4)
2 2 1 + 12 wh
where εa and εsf are permittivities of air and combination of 3D printed substrate and fluid in the
where εa and channel,
microfluidic εsf are permittivities
respectively [27].of air and combination of 3D printed substrate and fluid
The effectivechannel,
in the microfluidic permittivity of the combination
respectively [27]. of the inhomogeneous dielectric substrate can be
calculated using permittivity
The effective an equation forofeffective dielectric permittivity
the combination of the multilayered
of the inhomogeneous substrate,
dielectric [28]: can
substrate
be calculated using an equation for effective dielectric | | | permittivity
| | | of the multilayered substrate, [28]:
= ,
(5)
| d1 | + | d2 | + | d3 |
ε s f =      , (5)
where the coefficients dn are: d d 
 1+ 2 + 3
d 
εs ε s+ f εs

where the coefficients dn are:

= ln 2 ,
 r 
1
1+ 
πw

1  cosh 4h
1

d1 = ln 2 r ,
π  1− 1 

cosh 4h πw
= ln 2 1 , (6)
 s 
1
1+  
 cosh πw 
1  4( h1 + h c ) 
d2 = ln2
 ,
(6)
s
π  1−  1

= ln 2 .

πw
cosh
4( h1 + h c )
 s 
1+  1 
The εs+f is the dielectric constant of the
 middlecosh
layer4 2h
with
πw microchannel
 that can be calculated
1  ( 1 +hc ) 
using Bruggeman formalism [29]: d3 = ln2  s .

π  1−  1
= + cosh
1 − πw ,

(7)
4(2h1 +hc )

where V is the volumetric fraction of the microfluidic channel in the surrounding PLA substrate.
Based on the above equations, the operating principle of the sensor can be described. The change
of the fluid’s properties in the microfluidic channel causes the change of εs+f, which results in the
change of effective permittivity of the microstrip. Consequently, the phase velocity changes, which
Sensors 2017, 17, 892 6 of 14

The εs+f is the dielectric constant of the middle layer with microchannel that can be calculated
using Bruggeman formalism [29]:
ε s+ f = Vε s + (1 − V )ε f (7)

where V is the volumetric fraction of the microfluidic channel in the surrounding PLA substrate.
Based on the above equations, the operating principle of the sensor can be described. The change
of the fluid’s properties in the microfluidic channel causes the change of εs+f , which results in the change
of effective permittivity of the microstrip. Consequently, the phase velocity changes, which alters
the phase shift. It can be concluded that different values of the fluid’s permittivity is related to different
phase shifts, which is a necessary condition for constructing a calibration curve.
The defected EBG pattern in the sensor ground plane is used to improve the sensitivity
of the microstrip sensor. The introduction of the uniform EBG structure in the ground plane forms
a frequency region where propagation is forbidden, i.e., bandgap in the transmission characteristic [22].
The group velocity that can be determined by the slopes of the bands of the propagation modes
goes to zero at the bandgap edges in the case of the microstrip EBG sensors [21]. In that manner,
for a constant operating frequency, a significant change in the group velocity can be observed in the case
of the microstrip with EBG in comparison to the conventional microstrip line. A large decrease
in the group velocity corresponds to the slow-wave effect.
The defect in the EBG results in a resonance in the bandgap, which frequency is determined
by the size of the defect. By introducing defected EBG structure in the microstrip ground plane,
the phase change significantly increases, especially at the frequencies that are close to the bandgap
edges and at the resonance in the bandgap. As is shown in [21], at a constant frequency, the change
in wave vector, k is larger for the band with lower group velocity (∆k). The phase change (∆ϕ)
for a given change in permittivity is proportional to ∆k LTL , where LTL is the effective length
of the microstrip sensor. This is the main cause of increased phase change and therefore sensitivity
of the proposed sensor.
It can be mentioned that from the phase difference measurement, the real part of rgw dielectric
constant can be directly determined. The complex permittivity of the fluid can be obtained
by measuring both the amplitude and the phase of the transmitted signal of the sensor and incident
signal, or it can be reconstructed from Kramers–Kronig dispersion relations.
The phase-shift method allows characterization of a sample on single frequency, unlike resonant
methods that require characterization over a range of frequencies. In addition, phase-shift measurement
is less prone to the noise and less sensitive to the insertion loss. In that manner, this method is suitable
for sensing of the high-loss materials.

4. Simulation Results
The characteristics of the proposed sensor, the influence of different geometrical parameters,
optimization and the influence of the different fluids in the microchannel have been analysed using CST
Microwave studio. PLA is chosen as substrate material since it is one of the most used thermoplastics
in 3D printing. Initially, dialectic constant of the used 3D printed PLA material printed with 100%
infill was determined to be 2.7 with tanδ equal to 0.01 at the frequency of 6 GHz. The optimized
dimensions of the microstrip line, microfluidic channel and EBG structure have been determined
to be: h = 1.5 mm, lstrip = 89.84 mm, w1 = 1 mm, w = 3.6 mm, l2 = 20 mm, l1 = 100 mm, hc = 0.4 mm,
wc = 350 µm, a = 2.35 mm, b = 1.65 mm, c = 4.1 mm, dEBG = 8 mm, pEBG = 13.4 mm, and lend = 5.8 mm.
The simulation results of the proposed sensor with different fluids placed in the microchannel
are shown in Figure 4. Each fluid in the simulation is modelled with its material parameters, i.e.,
its permittivity and dissipation factor, as shown in Table 1 [30].
The EBG structure is designed to provide bandgap between 5 and 9 GHz, while the defect
in the EBG causes the resonant effect at 6 GHz. From the transmission characteristic, it can be seen that
the resonant frequency of the defect in the band gap slightly shifts due to the change of the dielectric
constant of the material in microchannel, Figure 4a. On the other hand, the phase change is increasing
Sensors 2017, 17, 892 7 of 14

due to a decrease in the wave phase velocity, Figure 4b. The effect of the EBG structure is predominant
at the frequency of 6 GHz where the wave phase velocity is minimal. This frequency is recognized
as a frequency of interest since the change of the phase difference is the highest in that case. It should
be noted that changes of the resonance and loss in the transmitted signal depend on the material
in the microfluidic channel. However, the insertion losses do not fall below—10 dB in the worst case
at operating frequency. Therefore, the phase of the transmitted signal can be measured using standard
phase comparators or detectors that determine phase difference as a subtraction of the phase 7shift
Sensors 2017, 17, 892 of 13
of the sensor transmitted signal and the phase of the excitation signal.
The
The simulation
simulation results
results show
show that
that the
the change
change of of the
the fluid
fluid permittivity
permittivity from
from 11 (air)
(air) to
to 80.1
80.1 (water)
(water)
causes the phase-shift difference of 84 ◦ . Compared to the phase shift of the conventional microstrip
causes the phase-shift difference of 84°. Compared to the phase shift of the conventional microstrip
line ◦
linewithout
withoutdefected
defectedEBG
EBGwhich
whichisisonly
only10.2
10.2° at
at 66 GHz,
GHz, the
the proposed
proposed design
design shows
shows eight
eight times
times higher
higher
phase shift.
phase shift.

(a) (b)
Figure4.4.Simulation
Figure Simulation results
results of
of the
the proposed
proposed sensor
sensor with
with different
different fluids
fluids placed
placed in
in the
the microchannel:
microchannel:
(a)transmission
(a) transmissioncharacteristic;
characteristic;(b)
(b)normalized
normalizedphase.
phase.

Table 1. Dielectric properties of analysed fluids at room temperature.

Table 1. Dielectric properties of analysed fluids at room temperature.
Fluid εr tanδ
Fluid
Toluene εr 2.3 tanδ
0.04
Toluene
Isopropyl 2.319.7 0.04
0.799
Isopropyl
Ethanol 19.724.5 0.799
0.941
Ethanol 24.5 0.941
Methanol 32 32
Methanol 0.659
0.659
WaterWater 80.180.1 0.123
0.123

5. Fabrication
5. Fabrication and
and Measurement
Measurement
Microfluidicchannel,
Microfluidic channel,embedded
embeddedinto intothe
thesubstrate,
substrate,was
wasdesigned
designedusing
usingthe
the3D3Dmodelling
modellingCAD CAD
software. The
software. The effects
effects that
that occur
occur during
during 3D 3D printing,
printing, such
such asas shrinkage
shrinkage and and variations
variations of of the
the final
final
dimensions, were taken into account in the final model to obtain designed dimensions
dimensions, were taken into account in the final model to obtain designed dimensions of the channel of the channel
andthickness
and thicknessofof
thethe substrate.
substrate. 3D3D model
model has has
beenbeen imported
imported into KISSlicer
into KISSlicer software
software to create tothe
create
G-codethe
G-code with the following slicing parameters: layer thickness of 0.1 mm, extrusion
with the following slicing parameters: layer thickness of 0.1 mm, extrusion width of 0.25 mm, number width of 0.25 mm,
number
of of perimeter
perimeter equal toequal to 4,factor
4, infill infill of
factor
100%,of 100%, andprint
and the the print
speed speed
of 20ofmm/s.
20 mm/s. InInorder
ordertotomake
make
aaleak-proof
leak-proof structure,
structure, the 3Dtheprinted
3D printed
model ismodel is using
designed designed using
full infill fulland
factor infill
highfactor and high
infill/perimeter
infill/perimeter overlapping
overlapping settings. settings.
The microfluidic
The microfluidicsensor’s
sensor’s substrate
substrate andand microfluidic
microfluidic channel
channel areareprinted
printedsimultaneously
simultaneously using using
aa Felix
Felix 3.1
3.1 3D
3Dprinter
printerbased
based onon FDMFDM technology.
technology. Biodegradable PLA thermoplastic
Biodegradable PLA thermoplastic filament with a
filament
diameter of 1.75 mm is used since it results in well-defined and quality printed
with a diameter of 1.75 mm is used since it results in well-defined and quality printed structures. structures. The
extruder temperature was set to 190 °C for the first layer, while 185 °C was used for other layers. It
should be noted that the designed microchannel is printed without any supporting material. Figure
5a shows the printing process of the 3D substrate with embedded microchannel.
The layout of the fabricated substrate and microfluidic channel filled with coloured fluid are
shown in Figures 5b and 6, respectively. The final dimensions of the fabricated microfluidic channel
are determined by measuring the horizontal and vertical cross sections of the fabricated prototypes
Sensors 2017, 17, 892 8 of 14

The extruder temperature was set to 190 ◦ C for the first layer, while 185 ◦ C was used for other
layers. It should be noted that the designed microchannel is printed without any supporting material.
Figure 5a shows the printing process of the 3D substrate with embedded microchannel.
The layout of the fabricated substrate and microfluidic channel filled with coloured fluid are
shown in Figures 5b and 6, respectively. The final dimensions of the fabricated microfluidic channel are
determined by measuring the horizontal and vertical cross sections of the fabricated prototypes using
the Huvitz HRM 300 profilometer, as shown in Figure 7. The measured dimensions of the fabricated
microfluidic channel and substrate are: h = 1.52 mm, l1 = 100.1 mm, hc = 367–383 µm, wc = 343–384 µm,
a = 2.15–2.71 mm, b = 1.45–1.72 mm, c = 4.02–4.23 mm. Imperfections of the 3D printing process mostly
affect
Sensors the
2017,dimensions
17, 892 of the channel causing the variation up to 10% from the designed values. 8 of 13
Sensors 2017, 17, 892 8 of 13
Sensors 2017, 17, 892 8 of 13

Figure 5.
Figure 5. Sensor
Sensor fabrication
fabrication process:
process: (a)
(a) Printing
Printing process
process of
of the 3D substrate
the 3D substrate with
with embedded
embedded
microchannel;
microchannel; (b) Layout of
(b)fabrication the 3D
Layout ofprocess: printed
the 3D (a)
printed substrate with
substrate embedded microfluidic channel; (c)
Figure 5. Sensor Printing processwith embedded
of the microfluidic
3D substrate channel;
with embedded
Figure
(c) 5. Sensor
Conductive
Conductivetoptop fabrication
and bottom process:
layers (a) Printing
precisely cut with process
laser, of (d)
and the 3D substrate
(d)Layout of
ofthe with embedded
the proposed sensors,
microchannel; (b) and bottom
Layout layers
of the 3Dprecisely
printed cut with
substratelaser,
withand Layout
embedded proposed
microfluidic sensors,
channel; (c)
microchannel;
top layer
top layer and (b) Layout
and bottom
bottom of the
layer with
layer with 3D printed
mounted
mounted SMA substrate
SMA with embedded microfluidic channel; (c)
connectors.
connectors.
Conductive top and bottom layers precisely cut with laser, and (d) Layout of the proposed sensors,
Conductive top and bottom layers precisely cut with laser, and (d) Layout of the proposed sensors,
top layer and bottom layer with mounted SMA connectors.
top layer and bottom layer with mounted SMA connectors.

Figure 6. Layout of the 3D printed substrate with embedded microfluidic channel filled with coloured
fluid.
Figure
Figure 6.
6.Layout of the
Layout 3D printed
of the substrate
3D printed with embedded
substrate microfluidic
with embedded channel channel
microfluidic filled with coloured
filled with
Figure 6. Layout of the 3D printed substrate with embedded microfluidic channel filled with coloured
fluid.
coloured fluid.
fluid.

Figure 7. Cross section of the 3D printed microfluidic channel with measured widths.
Figure 7. Cross section of the 3D printed microfluidic channel with measured widths.
The conductive
Figure
Figure 7.
partssection
7. Cross
in the of
Cross section
top and
of the
the 3D
bottommicrofluidic
3D printed
layers werechannel
printed microfluidic
realized
channel with
with
using 40 μm
measured
measured
thick conductive
widths.
widths.
aluminium sticky tape
The conductive partsprecisely
in the topcutandwith Rofin-Sinar
bottom PowerLine
layers were realizedDusing
laser,40Figure
μm thick5c. conductive
The sticky
The conductive parts in the top andand bottom layers were realized using 40 μm thick conductive
aluminium sticky tape precisely cut with Rofin-Sinar PowerLine D laser, Figure 5c. The layout
conductive tape is accurately positioned affixed from both sides of the substrate. The final sticky
aluminium sticky
of the proposed tape with
sensors precisely cut with
mounted Rofin-Sinar
end-lunch SMA PowerLineis D
connectors laser,inFigure 5c. The sticky
conductive tape is accurately positioned and affixed from both sides ofshown Figure
the substrate. 5d.final
The layout
conductive tape is accurately positioned and affixed8,from boththe
sides of the substrate.
are The final layout
of theThe measurement
proposed sensors setup is shownend-lunch
with mounted in Figure SMAwhere
connectors fluids injected
is shown in Figuretrue
5d. additional
of the proposed
tubules the sensors with mountedusing end-lunch SMA connectors is shown in Figure 5d.
Theinto microfluidic
measurement setup channel syringe.
is shown in Figure 8, The characteristic
where the fluids parameters
injected areoftruethe additional
fabricated
sensorThe measurement
were measured setup
in the is shown range
frequency in Figure 8, where
between 100 the and
kHz fluids
8 injected
GHz usingare true
two additional
ports Agilent
tubules into the microfluidic channel using syringe. The characteristic parameters of the fabricated
tubules into
8501C Vector the microfluidic
Network in channel
Analyser. using syringe. The characteristic parameters of the fabricated
sensor were measured the frequency range between 100 kHz and 8 GHz using two ports Agilent
Sensors 2017, 17, 892 9 of 14

The conductive parts in the top and bottom layers were realized using 40 µm thick conductive
aluminium sticky tape precisely cut with Rofin-Sinar PowerLine D laser, Figure 5c. The sticky
conductive tape is accurately positioned and affixed from both sides of the substrate. The final
layout of the proposed sensors with mounted end-lunch SMA connectors is shown in Figure 5d.
The measurement setup is shown in Figure 8, where the fluids injected are true additional tubules
into the microfluidic channel using syringe. The characteristic parameters of the fabricated sensor
were measured in the frequency range between 100 kHz and 8 GHz using two ports Agilent 8501C
Vector Network Analyser.
The simulation and the measurement results are compared in terms of transmitted amplitude
and phase, Figure 9. For better visibility, the measurements and simulations were compared
only for two fluids, i.e., fluids which have the lowest and the highest dielectric constant, i.e., air
and water. It can be seen that the simulated and measured results are in excellent agreement.
The frequency where the phase shift shows the highest changes is slightly shifted to 6.15 GHz due
to imperfection of the fabrication process, while the phase difference is slightly increased. On the other
hand, the insertion
Sensors 2017, 17, 892 losses do not fall below −10 dB in the worst case at the operating frequency.9 of 13

Sensors 2017, 17, 892 9 of 13

Figure 8. Measurement setup of the proposed sensor.

Figure 8. Measurement setup of the proposed sensor.
Figure 8. Measurement setup of the proposed sensor.

(a) (b)
Figure. 9. Comparison(a)
of the measured and simulated responses: (a) Transmission
(b) characteristic; and
(b) Normalized phase.
Figure.9.9.Comparison
Figure Comparisonofofthe
themeasured
measuredand
andsimulated
simulatedresponses:
responses:(a)
(a)Transmission
Transmissioncharacteristic;
characteristic;and
and
(b)
(b) Normalized
Normalized phase.
phase.
Figure 10 shows the measured transmission and phase characteristics of the proposed sensor
with different fluids in the microfluidic channel in the frequency range of interest. The empty air
Figure 10 shows the measured transmission and phase characteristics of the proposed sensor
channel is used
Figure as a reference
10 shows value. For
the measured better visibility,
transmission the phase
and phase characteristic
characteristics is normalized
of the in the
proposed sensor
with different fluids in the microfluidic channel in the frequency range of interest. The empty air
range from −180° to 180°.
with different fluids in the microfluidic channel in the frequency range of interest. The empty air
channel is used as a reference value. For better visibility, the phase characteristic is normalized in the
channel is used as a reference value. For better visibility, the phase characteristic is normalized
range from −180° to 180°.
in the range from −180◦ to 180◦ .
 (b) Normalized phase.

Figure 10 shows the measured transmission and phase characteristics of the proposed sensor
with different fluids in the microfluidic channel in the frequency range of interest. The empty air
channel is used
Sensors 2017, 17, 892as a reference value. For better visibility, the phase characteristic is normalized 10
inofthe
14
range from −180° to 180°.

(a) (b)
Figure 10. Measured
Figure 10. Measuredresults
results
of of
thethe proposed
proposed sensor
sensor with with different
different fluids the
fluids inside inside the microfluidic
microfluidic channel:
channel: (a) Transmission characteristic; and (b) Normalized
(a) Transmission characteristic; and (b) Normalized phase. phase.

For more comprehensive analysis, Table 2 summarizes the phase shifts for different fluids in the
For more comprehensive analysis, Table 2 summarizes the phase shifts for different fluids
microfluidic channel obtained by simulations and measurements, as well as the phase shifts
in the microfluidic channel obtained by simulations and measurements, as well as the phase
calculated based on the above-stated equations, where εsf denotes effective permittivity of a
shifts calculated based on the above-stated equations, where ε denotes effective permittivity
combination of 3D printed substrate and fluid in the microfluidicsfchannel, εeff is the total effective
of a combination of 3D printed substrate and fluid in the microfluidic channel, εeff is the total effective
permittivity, while ∆ϕcal , ∆ϕsim and ∆ϕmeas are calculated, simulated and measured phase shifts,
respectively. The calculated and simulated results agree well with the measured results.

Table 2. Calculated, simulated and measured phase shifts for different fluids in the microchannel.

Fluid εsf εeff ∆ϕcal ∆ϕsim ∆ϕmeas

Air 2.6946 1.5014 0 0 0
Toluene 2.6987 1.5026 17.02 16.39 17.7
Isopropyl 2.7518 1.5183 48.45 58.05 47.5
Ethanol 2.7657 1.5224 62.13 74.47 55.9
Methanol 2.7868 1.5287 76.12 79.12 71.7
Water 2.9066 1.5641 85.75 84 86.7

The sensitivity of the proposed sensor can be defined as a ratio of the phase difference of sensor
response with fluid and air in the microchannel, divided by permittivity of used fluid. Figure 11
shows the phase-shift change of the proposed microfluidic EBG sensor with respect to the change
of permittivity of the fluid. For a better comparison, the results of the conventional microstrip line
sensor, as well as results obtained from simulations, are added in Figure 11. The exponential fitting
curves and corresponding equations that provide excellent curve-fitting are also presented in Figure 11.
The Curve Fitting Tool in the Matlab was used. It can be seen that the proposed sensor shows
relatively high and almost linear dependence for the fluid materials with permittivity lower than 30.
For the higher values of the permittivity, the change in phase is relatively small and sensor goes
to saturation. It can be seen that in the case of the conventional microstrip line, eight times lower
sensitivity is achieved while the saturation occurs for permittivity of 20. Furthermore, the measurement
results show the better linearity in comparison with the simulations, especially for the lower values
of dielectric constant.
 relatively high and almost linear dependence for the fluid materials with permittivity lower than 30.
For the higher values of the permittivity, the change in phase is relatively small and sensor goes to
saturation. It can be seen that in the case of the conventional microstrip line, eight times lower
sensitivity is achieved while the saturation occurs for permittivity of 20. Furthermore, the
measurement results show the better linearity in comparison with the simulations, especially 11
Sensors 2017, 17, 892 forofthe
14
lower values of dielectric constant.

Figure11.
Figure 11.Sensitivity
Sensitivityof
ofthe
the proposed
proposedEBGEBGsensor
sensorcompared
comparedtotothe
theconventional
conventionalmicrostrip
microstripline
linesensor
sensor
(dots) and the corresponding fitting curves (lines) with equations.
(dots) and the corresponding fitting curves (lines) with equations.

The potential application is demonstrated through the implementation of proposed sensors for
The potential application is demonstrated through the implementation of proposed sensors
the detection of toluene concentration in toluene–methanol mixture, where various concentrations of
for the detection of toluene concentration in toluene–methanol mixture, where various concentrations
toluene were analysed. Figure 12 illustrates the measured phase difference when the toluene
of toluene were analysed. Figure 12 illustrates the measured phase difference when the toluene
concentration varies from 0% to 100%. The linear and polynomial fitting curves and corresponding
concentration varies from 0% to 100%. The linear and polynomial fitting curves and corresponding
equations that better describe the measured phase dependence on the concentration of toluene in the
equations that better describe the measured phase dependence on the concentration of toluene
toluene–methanol mixture are also shown in Figure 12. The experimental results show the phase
in the toluene–methanol mixture are also shown in Figure 12. The experimental results show the
difference changes for 54 degrees when the concentration of toluene is changed from 0% to 100%. In
phase difference changes for 54 degrees when the concentration of toluene is changed from 0% to 100%.
addition, the proposed sensor shows almost linear shift with sensitivity of 0.540° per percentage of
In addition, the proposed sensor shows almost linear shift with sensitivity of 0.540◦ per percentage
toluene. These results successfully demonstrate and confirm the application of the proposed
of toluene. These results successfully demonstrate and confirm the application of the proposed
microfluidic EBG sensor as chemical sensor.
Sensors 2017, 17,EBG
microfluidic 892 sensor as chemical sensor. 11 of 13

Figure 12. Measurement

Measurement of the toluene concentration in toluene–methanol mixture: sensor measured
response (dots) and the corresponding fitting curves (lines)
(lines) with
with equations.
equations.

The
The proposed
proposed sensor
sensor is compared with
is compared with recently
recently published
published microfluidic
microfluidic sensors
sensors that operate
that operate
according to the phase-shift method [31–34]. The parameters for these sensors are
according to the phase-shift method [31–34]. The parameters for these sensors are summarized summarized in
Table 3, where fopr denotes sensor operating frequency and Δφmax is the maximal phase shift. Table 3
in Table 3, where fopr denotes sensor operating frequency and ∆ϕmax is the maximal phase shift. Table 3
also
also shows
shows specific application of
specific application of the
the compared
compared sensors
sensors and
and used
used fabrication
fabrication technologies.
technologies.
The
The proposed sensor has comparable characteristics and its fabrication process is
proposed sensor has comparable characteristics and its fabrication process is the simplest
the simplest
and
and the cheapest compared
the cheapest compared to to other
other published
published sensors.
sensors. Although
Although one
one may
may argue
argue that
that the
the sensor
sensor
published in [34] has the largest phase shift, it should be noted that this sensor requires the most
demanding fabrication process. Compared with the sensor proposed in [31], which fabrication
complexity is fairly similar, the proposed EBG sensor shows 65% better sensitivity. Therefore, the
proposed design represents a good candidate for the design of a high-performance sensor since it
reconciles the requirements for good sensitivity, compactness and simple fabrication.
Sensors 2017, 17, 892 12 of 14

published in [34] has the largest phase shift, it should be noted that this sensor requires the most
demanding fabrication process. Compared with the sensor proposed in [31], which fabrication
complexity is fairly similar, the proposed EBG sensor shows 65% better sensitivity. Therefore,
the proposed design represents a good candidate for the design of a high-performance sensor since
it reconciles the requirements for good sensitivity, compactness and simple fabrication.

Table 3. Comparison of the characteristics of the proposed sensor and other recently published sensors
that operate according to the phase-shift method.

Ref. fopr ∆ϕmax [deg] Fabrication Technologies Applications

Polymethyl methacrylate (PMMA) +
micromachining Coplanar waveguide,
[31] 900 MHz 52 (εr = 1–80) CPW (composite right-left handed General microfluidic
transmission line, CRLH TL with
shorted stub)
CPW with Polydimethylsiloxane (PDMS)
[32] 1.84 GHz 94 (εr = 1–80) microfluidic channel on top of glass General microfluidic
wafer Borofloat33
Silicon-on-insulator (SOI) wafer with Red blood cell
[33] 1 Hz–10 MHz 90 (εr = 1–80)
Cr/Au electrodes, PDMS fluidic characterization
RF micro-electro-mechanical system
(MEMS), chemical vapor deposition CWP phase shifter
136 (at 6 GHz)
[34] 2–10 GHz (CVD) Parylene, Parylene with integrated
(εr = 1:36)
surface-micromachining, Silicon, micropumps
Cr/Au electrodes
General
This work 6 GHz 86 (εr = 1–80) 3D printing Microstrip microfluidic/Toluene
sensor

6. Conclusions
A novel microfluidic microstrip EBG sensor realized using simple 3D printing process has been
proposed in this article. The proposed sensor is composed of the microstrip line, defected EBG structure
etched in the ground plane, and microfluidic channel embedded in the microstrip substrate. The EBG
structure with single defect placed beneath the channel is used as a pattern in the ground plane
to improve sensor sensitivity. The microfluidic channel is fabricated using conventional 3D printing
technique without any supporting material simultaneously with the microstrip substrate.
The operating principle of the sensor is based on phase-shift measurement of the propagation
signal at single operating frequency. When a fluid flows in the microfluidic channel, the phase
of the propagating signal changes due to the different permittivity of the fluid. The sensor dimensions
were optimized using electromagnetic simulations while performances of the proposed sensor were
validated by measuring the phase response for different fluids in the microchannel.
The measurement results of the fabricated sensor show that the change of the permittivity of fluids
in the microchannel from 1 to 80 results in the phase shift of 86◦ . Moreover, the proposed sensor
shows relatively high and almost linear sensitivity for fluids which dialectic constant is lower than 30.
The potential application is demonstrated by the implementation of a proposed sensor for the detection
of toluene concentration in toluene–methanol mixture. The experimental results show that the phase
difference linearly changes when the concentration of toluene changes from 0% to 100% with sensitivity
of 0.54◦ per percentage of toluene.
In this paper, we propose a novel low-cost, reusable, and easily fabricated design that uses small
volumes of fluid. The proposed sensor is characterized with relatively high sensitivity and linearity,
which makes it a suitable candidate for monitoring small concentrations of specific fluid in different
mixtures. For future consideration, we are planning to design supporting electronics, to minimize
the sensor dimensions, and improve its sensitivity. The sensor will be further tested on various fluids
and mixtures used in biomedical applications and industry.
Sensors 2017, 17, 892 13 of 14

Acknowledgments: This work has been funded by FP7-REGPOT INNOSENSE GA No. 316191 (Reinforcement of
BioSense Center—ICT for Sustainability and Eco-Innovation).
Author Contributions: The authors contributed equally to this work. V.R. proposed the idea, performed
simulations and optimization, carried out the experiments and data analysis, and prepared the manuscript drafts.
G.K. contributed ideas, method for characterization, and contributed to the manuscript drafts. S.B. manufactured
the sensors, performed experiment, as well as contributed to the manuscript drafts.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Li, D. Encyclopedia of Microfluidics and Nanofluidics, 3rd ed.; Springer: New York, NY, USA, 2015.
2. Tabeling, P. Introduction to Microfluidics; Oxford University Press: Oxford, UK, 2005.
3. Kakaç, S.; Kosoy, B.; Li, D.; Pramuanjaroenkij, A. Microfluidics Based Microsystems—Fundamentals and
Applications; Springer: Rotterdam, The Netherlands, 2010.
4. Lin, B. Microfluidics—Technologies and Applications; Springer: Berlin/Heidelberg, Germany, 2011.
5. Minteer, S.D. Microfluidic Techniques; Humana Press: Totowa, NJ, USA, 2006.
6. Bogue, R. 3D printing: An emerging technology for sensor fabrication. Sens. Rev. 2016, 36, 333–338.
[CrossRef]
7. Bhattacharjee, N.; Urrios, A.; Kanga, S.; Folch, A. The upcoming 3D-printing revolution in microfluidics.
Lab Chip 2016, 16, 1720–1742. [CrossRef] [PubMed]
8. Ho, C.M.B.; Ng, S.H.; Lia, K.H.H.; Yoon, Y.-J. 3D printed microfluidics for biological applications. Lab Chip
2015, 15, 3627–3637. [CrossRef] [PubMed]
9. Capel, A.J.; Edmondson, S.; Christie, S.D.R.; Goodridge, R.D.; Bibb, R.J.; Thurstans, M. Design and additive
manufacture for flow chemistry. Lab Chip 2013, 13, 4583–4590. [CrossRef] [PubMed]
10. Symes, M.D.; Kitson, P.J.; Yan, J.; Richmond, C.J.; Cooper, G.J.T.; Bowman, R.W.; Vilbrandt, T.; Cronin, L.
Integrated 3D-printed reactionware for chemical synthesis and analysis. Nat. Chem. 2012, 4, 349–354.
[CrossRef] [PubMed]
11. Bishop, G.W.; Satterwhite, J.E.; Bhakta, S.; Kadimisetty, K.; Gillette, K.M.; Chen, E.; Rusling, J.F. 3D-printed
fluidic devices for nanoparticle preparation and flow-injection amperometry using integrated Prussian blue
nanoparticle-modified electrodes. Anal. Chem. 2015, 87, 5437–5443. [CrossRef] [PubMed]
12. Kadimisetty, K.; Mosa, I.M.; Malla, S.; Satterwhite-Warden, J.E.; Kuhns, T.M.; Faria, R.C.;
Lee, N.H.; Rusling, J.F. 3D-printed supercapacitor-powered electrochemiluminescent protein immunoarray.
Biosens. Bioelectron. 2016, 77, 188–193. [CrossRef] [PubMed]
13. Johnson, B.N.; Lancaster, K.Z.; Hogue, I.B.; Meng, F.; Kong, Y.L.; Enquist, L.W.; McAlpine, M.C. 3D printed
nervous system on a chip. Lab Chip 2016, 16, 1393–1400. [CrossRef] [PubMed]
14. Cook, B.S.; Cooper, J.R.; Tentzeris, M.M. An inkjet-printed microfluidic RFID-enabled platform for wireless
lab-on-chip applications. IEEE Trans. Microw. Theory Tech. 2013, 61, 4714–4723. [CrossRef]
15. Potyrailo, R.A.; Morris, W.G. Multianalyte chemical identification and quantitation using a single radio
frequency identification sensor. Anal. Chem. 2007, 79, 45–51. [CrossRef] [PubMed]
16. Mateu, J.; Orloff, N.; Rinehart, M.; Booth, J.C. Broadband permittivity of liquids extracted from transmission
line measurements of microfluidic channels. In Proceedings of the 2007 IEEE/MTT-S International
Microwave Symposium, Honolulu, HI, USA, 3–8 June 2007.
17. Chretiennot, T.; Dubuc, D.; Grenier, K. A microwave and microfluidic planar resonator for efficient and
accurate complex permittivity characterization of aqueous solutions. IEEE Trans. Microw. Theory Tech. 2013,
61, 972–978. [CrossRef]
18. Hayes, G.J.; So, J.-H.; Qusba, A.; Dickey, M.D.; Lazzi, G. Flexible liquid metal alloy (EGaIn) microstrip patch
antenna. IEEE Trans. Antennas Propag. 2012, 60, 2151–2156. [CrossRef]
19. Salim, A.; Lim, S. Complementary Split-Ring Resonator-Loaded Microfluidic Ethanol Chemical Sensor.
Sensors 2016, 16, 1802. [CrossRef] [PubMed]
20. Huang, P.-C.; Wang, M.-H.; Chen, M.-K.; Jang, L.-S. Experimental analysis of time-phase-shift flow sensing
based on a piezoelectric peristaltic micropump. J. Phys. D Appl. Phys. 2016, 49, 175402. [CrossRef]
21. García-Baños, B.; Cuesta-Soto, F.; Griol, A.; Catalá-Civera, J.M.; Pitarch, J. Enhancement of sensitivity
of microwave planar sensors with EBG structures. IEEE Sens. J. 2006, 6, 1518–1522. [CrossRef]
Sensors 2017, 17, 892 14 of 14

22. Griol, A.; Mira, D.; Martinez, A.; Marti, J. Multiple frequency photonic bandgap microstrip structures based
on defects insertion. Microw. Opt. Technol. Lett. 2003, 36, 479–481. [CrossRef]
23. Chang, Y.M.; Lee, J.C.; Chen, J.R.; Liaw, H.J.; Shu, C.M. Flammability characteristics studies on toluene and
methanol mixtures with different vapor mixing ratios at 1 atm and 150 ◦ C. J. Therm. Anal. Calorim. 2008, 93,
183–188. [CrossRef]
24. Poon, R.; Chu, I.; Bjarnason, S.; Potvin, M.; Vincent, R.; Miller, R.B.; Valli, V.E. Inhalation toxicity study of
methanol, toluene, and methanol/toluene mixtures in rats: Effects of 28-day exposure. Toxicol. Ind. Health
1994, 10, 231–245. [CrossRef] [PubMed]
25. Balanis, C.A. Advanced Engineering Electromagnetics; John Wiley & Sons Publisher Inc.: New York, NY,
USA,1989.
26. Abdelgwad, A.H.; Said, T.M.; Gody, A.M. Microwave detection of water pollution in underground pipelines.
Int. J. Wirel. Microw. Technol. 2014, 3, 1–15. [CrossRef]
27. Liu, R.; Zhang, Z.; Zhong, R.; Chen, X.; Li, J. Nanotechnology Synthesis Study: Technical Report 0-5239-1;
Department of Electrical and Computer Engineering, University of Houston: Houston, TX, USA, 2007;
pp. 76–79.
28. Jha, K.R.; Singh, G. Terahertz Planar Antennas for Next Generation Communication; Springer International
Publishing: Cham, Switzerland, 2014.
29. Lakhtakia, A.; Michel, B.; Weiglhofer, W.S. Bruggeman formalism for two models of uniaxial composite
media; dielectric properties. Compos. Sci. Technol. 1997, 57, 185–196. [CrossRef]
30. Kappe, C.O.; Dallinger, D.; Murphree, S.S. Practical Microwave Synthesis for Organic Chemists, 1st ed.;
Wiley-VCH: Weinhem, Germany, 2009.
31. Choi, S.; Su, W.; Tentzeris, M.M.; Lim, S. A novel fluid-reconfigurable advanced and delayed phase line using
inkjet-printed microfluidic composite right/left-handed transmission line. IEEE Microw. Wirel. Compon. Lett.
2015, 25, 142–144. [CrossRef]
32. Murray, C.; Franklin, R.R. Design and characterization of microfluidic housing effects on coplanar waveguide
microfluidic delay line performance. Microw. Opt. Technol. Lett. 2013, 55, 789–793. [CrossRef]
33. Cho, Y.H.; Yamamoto, T.; Sakai, Y.; Fujii, T.; Kim, B. Development of microfluidic device for electrical/physical
characterization of single cell. J. Mech. Syst. 2006, 15, 287–295. [CrossRef]
34. Tang, H.; Donnan, R.; Parini, C. Phase shifting with coplanar transmission line integrated electrostatic peristaltic
micropumps. In Proceedings of the Radar Conference EURAD 2005, Paris, France, 3–4 October 2005.

© 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/).