chemosensors

Article
Resistance-Capacitance Gas Sensor Based on
Fractal Geometry
Taicong Yang 1 , Fengchun Tian 1, * , James A. Covington 2, * , Feng Xu 1 , Yi Xu 3 , Anyan Jiang 1 ,
Junhui Qian 1 , Ran Liu 1 , Zichen Wang 4 and Yangfan Huang 1
1 College of Communication Engineering, Chongqing University, 174 Sha Pingba, Chongqing 400044, China
2 School of Engineering, University of Warwick, Coventry CV4 7AL, UK
3 College of Optoelectronic Engineering, Chongqing University, 174 Sha Pingba, Chongqing 400044, China
4 Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA
* Correspondence: FengchunTian@cqu.edu.cn (F.T.); J.A.Covington@warwick.ac.uk (J.A.C.);
Tel.: +86-23-6511-1745 (F.T.); +44-024-7657-4494 (J.A.C.)




Received: 25 April 2019; Accepted: 8 July 2019; Published: 15 July 2019


Abstract: An important component of any chemiresistive gas sensor is the way in which the resistance
of the sensing film is interrogated. The geometrical structure of an electrode can enhance the
performance of a gas-sensing device and in particular the performance of sensing films with large
surface areas, such as carbon nanotubes. In this study, we investigated the influence of geometrical
structure on the performance of gas sensors, combining the characteristics of carbon nanotubes with
a novel gas sensor electrode structure based on fractal geometry. The fabricated sensors were tested
with exposure to nitric oxide, measuring both the sensor resistance and capacitance (RC) of the
sensor responses. Experimental results showed that the sensors with fractal electrode structures
had a superior performance over sensors with traditional geometrical structures. Moreover, the RC
characteristics of these fractal sensors could be further improved by using different test frequencies
that could aid in the identification and quantification of a target gas.

Keywords: gas sensor; fractal; carbon nanotubes; RC characteristics

1. Introduction
Electronic noses (e-nose) are odor-recognizing instruments that consist of an array of gas sensors
and pattern recognition algorithm. These do not attempt to identify individual chemicals but analyze
the sample as a whole, allowing it to chemically sense complex odors that might be challenging
when using other techniques. These instruments can be made cost-effective, tailored to a specific
application and, where critical, can be of a non-invasive/non-destructive nature. They have been
used in a broad range of applications, including medical diagnostics, food manufacture, industrial
production, agriculture, and aerospace [1–3]. The gas sensor within the e-nose is one of the most critical
components for sensing odors, which is a bottleneck for the sensitivity, precision, and response time of
these instruments.
When the gas/odor molecule comes into contact with the sensitive material of the sensor, the gas
molecules react chemically. Then the sensor converts the chemical signal into a form than can be
detected as an electrical signal. Though there are a range of different gas/odor sensors, there are only
a few companies which sell chemiresistive sensors that employ a spectrum of detection approaches.
This is because many chemiresistive gas sensors do not respond well at low concentrations and have
slow response times. Therefore, improving its sensitivity and speeding up the sensor response have
become major research focuses [4,5]. A key factor affecting the performance of the sensor is the means
of interrogating the sensing material. The influence of a specific surface area on sensitivity, is much

Chemosensors 2019, 7, 0031; doi:10.3390/chemosensors7030031 www.mdpi.com/journal/chemosensors

Chemosensors 2019,
Chemosensors 2019, 7,
6, 0031
x 22 of 10
of 11

greater than that of the material itself [6]. Therefore, increasing a specific surface area [7] and
improving
greater thanthethatstructure of sensor
of the material are[6].
itself beneficial
Therefore, to improving
increasing athe sensitivity
specific surfaceof area
a sensor [8].improving
[7] and However,
most
the studiesofhad
structure adopted
sensor traditional
are beneficial Euclideanthe
to improving geometry forof
sensitivity sensor
a sensorshape, with simple
[8]. However, mostelectrode
studies
structures
had adopted that do not interrogate
traditional Euclidean all of the sensing
geometry for sensor film and with
shape, are typically formed of
simple electrode two parallel
structures that
electrodes
do [9,10]. Later
not interrogate all ofdesigns
the sensing and the
filmmajority
and are of MEMSformed
typically (micro-electro-mechanical-systems)
of two parallel electrodes [9,10]. gas
sensors
Later that employ
designs and themicro-hot
majority of plates,
MEMS operated interdigitated fingers to try and
(micro-electro-mechanical-systems) gas increase the volume
sensors that employ
of the sensing
micro-hot plates,film interrogated
operated and reduce
interdigitated the measured
fingers to try and resistance
increase the of volume
the sensors [11].
of the However,
sensing film
with optimization,
interrogated and reduce further improvements
the measured in the
resistance sensing
of the sensors film
[11].interrogation
However, with (and potentially
optimization,
sensitivity)
further could be achieved
improvements through
in the sensing filmmore sophisticated
interrogation (andelectrode
potentially structures.
sensitivity) could be achieved
Themore
through term sophisticated
“fractal” was electrode
first proposed in 1975 by Benoit Mondelbrot. He was both a researcher
structures.
at theThe
Physics
term Department
“fractal” wasoffirst the proposed
IBM Research in 1975Center (United
by Benoit States) andHe
Mondelbrot. a Professor
was bothofa Mathematics
researcher at
at Harvard
the University. For
Physics Department more
of the IBMthan 30 years
Research fractal
Center theory
(United has influenced
States) and a Professorvarious of disciplines
Mathematics andat
have been
Harvard applied inFor
University. biology,
more geophysics,
than 30 years physics,
fractal chemistry, astronomy,various
theory has influenced materials science, and
disciplines and other
have
fields.applied
been Many traditional
in biology, scientific
geophysics, problems
physics,have been solved
chemistry, by thematerials
astronomy, introduction of fractal
science, theory.
and other As
fields.
a newtraditional
Many concept and method,
scientific fractal is
problems being
have beenapplied
solvedandby theexplored in many
introduction fields. theory.
of fractal The well-known
As a new
American
concept andphysicist
method,Wheeler
fractal issaid:
beingWho is notand
applied familiar with in
explored fractals
manyin the future,
fields. who cannotAmerican
The well-known be called
a scientific
physicist cultural
Wheeler manWho
said: [12].is not familiar with fractals in the future, who cannot be called a scientific
Here,
cultural man we[12].
report on the design of a new gas sensor structure based on fractal geometry. This
sensor haswe
Here, a dual
report capacitance
on the design andofresistance
a new gasread sensoroutstructure
that are based
basedon onfractal
fractalgeometry.
geometry.This Thesensor
basic
circuit
has model
a dual for this sensor
capacitance has also read
and resistance been outanalyzed.
that areSincebasedcarbon nanotubes
on fractal geometry.(CNTs)Theare good
basic gas
circuit
sensingfor
model materials
this sensorwithhasa fast
alsoresponse-time,
been analyzed.good Sincerepeatability,
carbon nanotubes are able(CNTs)
work at areroom
goodtemperature,
gas sensing
have stable
materials withphysical and chemical properties,
a fast response-time, and areare
good repeatability, easy to work
able process in sensor
at room manufacturing
temperature, [13–
have stable
18], we chose
physical it to be the
and chemical sensing material
properties, and are easyin our experiment,
to process to verify
in sensor the effectiveness
manufacturing [13–18],ofwe
ourchose
sensorit
structure.
to be the sensing material in our experiment, to verify the effectiveness of our sensor structure.

2. Fractal Electrode
2. Fractal Electrode Design
Design
In the field
field of
of geometry,
geometry, fractal
fractal breaks
breaks the
the barrier
barrier of
of traditional
traditional Euclidean
Euclidean geometry.
geometry. When it
comes to aa dimension,
dimension, we always intuitively think of a line or
or a curve as a typical one-dimensional
a curve as a typical one-dimensional
object and a plane
plane as
as aa typical
typical two-dimensional
two-dimensional object.
object. Studies by Giuseppe
Giuseppe Peano (1858–1932)
(1858–1932) [19]
[19]
and David Hilbert (1862–1943)
(1862–1943) [20] [20] have shown that a line segment can also be two-dimensional and
can fill
fill the
the entire
entire plane;
plane; as as illustrated
illustrated in
in Figure
Figure 1.
1.

Figure 1. Space-filling curve explored by Hilbert (First six steps/orders

steps/orders shown).
shown).
Chemosensors 2019,
Chemosensors 2019, 7,
6, 0031
x 33 of 10
of 11
Chemosensors 2019, 6, x 3 of 10
This Hilbert curve, which is based on fractal geometry theory, provides new ideas for the design
and This Hilbert curve,
manufacture of highwhich is based on
performance fractal
ongasfractal geometry
due theory,
geometry
sensors, theory,
to provides
its special new ideas
structure andfor the design
dimensions,
and manufacture
manufacture
compared of high performance
to conventional performance gas
gas sensors,
Euclidean geometry. sensors,Fordue
due to
to its
its special
capacitive special structure
structurethe
gas sensors, and
and dimensions,
dimensions,
special fractal
compared to conventional
dimensiontoprovides
conventional Euclidean
Euclidean
a theoretical geometry.
geometry.
basis For capacitive
For capacitiveof
for the manufacture gas gas sensors,
sensors, the
ultra-large the special
special fractal
capacitors [21]. On fractal
dimension
the one
dimension
provides
hand, electricprovides
a theoretical abasis
field linestheoretical
for thebasis
between the for plates
manufacture
two the manufacture
ofof
ultra-large of ultra-large
capacitors
the capacitor (shown capacitors
[21].
in On the 2a),
Figure [21].
one On the
hand,
represent one
electric
the
hand,
field electric
lines field
between lines
the two between
plates the
of the two plates
capacitor of the
(shown capacitor
in Figure (shown
2a), in
represent Figure
charge density and as the insulating layer in the middle of the parallel plate of Euclidean capacitor is the 2a),
charge represent
density the
and
charge
narrow, density
as the insulating and as in
layer
the insulating thetheinsulating
layermiddle oflayer
is easily in the middle
thebroken
parallel plate of
down ofEuclidean
when the the
parallel platefield
capacitor
electric of
is Euclidean
narrow, thecapacitor
is strong, insulating
therefore, is
narrow, the insulating
layer is easily
microelectronic broken down
measurementlayer is easily
when
cannotthe be broken
electric down
field
performed when
iswell.
strong, theother
On therefore,
the electric field is also
microelectronic
hand, there strong, therefore,
measurement
exists electric
microelectronic
field
cannotpenetration measurement
be performedin the lateral
well. cannot
Onspace be hand,
between
the other performed
the same
there well.
side
also On theelectric
plates
exists other
(shownhand,
field there also
2b). exists
in penetration
Figure Ifinthe electric
theparallel
lateral
field
spacepenetration
plates between
are paired in
the the lateral
same
laterally, side
theyspace
will between
plates (shown
store more the same side
in energy
Figure 2b). plates
and If the
will (shown
parallelainplates
produce Figure
bigger are2b). If thelaterally,
paired
capacitance. parallel
plates arestore
they will paired laterally,
more energythey andwill
willstore morea bigger
produce energycapacitance.
and will produce a bigger capacitance.
+++++++++
+++++++++ (a )
--------- (a )
---------
++++ ----
++++ ---- (b )
(b )
- - - - ++++
- - - - ++++
Figure 2. Illustration of the electrical field between capacitor plates.
Figure 2.
Figure Illustration of
2. Illustration of the
the electrical
electrical field
field between
between capacitor
capacitor plates.
plates.
Here, the capacitor plate was designed on a fractal structure, as shown in Figure 3. With this
Here, the capacitor plate was designed on a fractal structure, as shown in Figure 3. With this design,
Here,
design, thethe capacitor
special plate was
interlocking designed
edge on a fractal
can withstand structure,
a larger as shown
electric field andin Figure 3. With
generate this
a larger
the special interlocking edge can withstand a larger electric field and generate a larger capacitance.
design, the special interlocking edge can withstand a larger electric field and
capacitance. Whilst the flux of the transverse electric field depends on the edge length of the plate, generate a larger
Whilst the flux of the transverse electric field depends on the edge length of the plate, the design of the
capacitance.
the design ofWhilst the plate
the fractal flux of the
can transverse
greatly electric
increase field depends
the length on under
of the plate the edge thelength of the
same area, plate,
thereby
fractal plate can greatly increase the length of the plate under the same area, thereby resulting in the
the designin
resulting of the
the fractal
increase plate can electric
of the greatly increase the length
field strength and of the plate under
capacitance. The theareasame area, thereby
occupied by the
increase of the electric field strength and capacitance. The area occupied by the capacitor plates under
resulting in the under
capacitor plates increase of thefractal
different electric field control
orders strength andsame,
is the capacitance. The area
as the number occupied
of fractal by gets
orders the
different fractal orders control is the same, as the number of fractal orders gets larger, the ratio of plate
capacitor
larger, theplates
ratio under different
of plate length fractal
to the orders
occupied control
areaisgets
the bigger,
same, asasthe number
shown in of fractal
Figure orders
4 [21]. gets
Hirad
length to the occupied area gets bigger, as shown in Figure 4 [21]. Hirad Samavati and his coworkers
larger,
Samavatitheandratiohis
of coworkers
plate lengthshowed
to the occupied
that witharea gets bigger,
a horizontal as shown
spacing of 0.6in μm,
Figure 4 [21]. Hirad
a physical (not
showed that with a horizontal spacing of 0.6 µm, a physical (not simulated) fabricated fractal plate
Samavati
simulated)and his coworkers
fabricated fractal showed
plate withthatboundary
with a horizontal
dimensionspacing
of about of 0.6
1.6 μm,
has aover
physical
twice (not
the
with boundary dimension of about 1.6 has over twice the capacitance of a Euclidean capacitor of
simulated)
capacitance fabricated fractalcapacitor
of a Euclidean plate with
of the boundary
same area.dimension of about
Simulations show 1.6 thathasan over twice
increase of the
the same area. Simulations show that an increase of the boundary dimension of 1.80 would give a
capacitance of a Euclidean
boundary dimension of 1.80 capacitor
would give of athe same area.
capacitance fiveSimulations
times that ofshow that an increase
the Euclidean capacitorof[21].
the
capacitance five times that of the Euclidean capacitor [21]. As the capacitance increases, changes in gas
boundary dimensionincreases,
As the capacitance of 1.80 would givein
changes a capacitance five times
gas concentration canthat of the
result in Euclidean
larger sensor capacitor [21].
responses,
concentration can result in larger sensor responses, which in turn increases sensor sensitivity.
As theincapacitance
which turn increasesincreases,
sensor changes in gas concentration can result in larger sensor responses,
sensitivity.
which in turn increases sensor sensitivity.

Figure 3. Schematic of a fractal capacitor.

Figure 3. Schematic of a fractal capacitor.
Chemosensors 2019, 7, 0031 4 of 11
Chemosensors 2019, 6, x 4 of 10
Chemosensors 2019, 6, x 4 of 10

Figure 4. Ratio of plate length versus area under different fractal orders.
Ratio of
Figure 4. Ratio of plate
plate length
length versus
versus area
area under different fractal orders.

The
The parasitic
parasitic capacitance
capacitance of of the
the sensor
sensor refers
refers to
to the
the parasitic/stray
parasitic/stray capacitance
capacitance between
between the the plate
plate
The parasitic capacitance of the sensor refers to the parasitic/stray capacitance between the plate
and
and the
the substrate.
substrate. It
It not
not only
only changes
changes the
the capacitance
capacitance value
value of
of the
the sensor,
sensor, but
but also
also causes
causes the
the sensor
sensor
and the substrate. It not only changes the capacitance value of the sensor, but also causes the sensor
characteristics
characteristics to
to be
be unstable,
unstable, which
which can
can cause
cause serious
serious interference
interference asas the
the sensor
sensor itself
itself has
has aaa small
small
characteristics to be unstable, which can cause serious interference as the sensor itself has small
capacitance
capacitance and
and the
the parasitic
parasitic capacitance
capacitance is
is extremely
extremely unstable.
unstable. Especially
Especially when
when gas
gas concentration
concentration
capacitance and the parasitic capacitance is extremely unstable. Especially when gas concentration
rapidly
rapidly changes,
changes, the
the influence
influence ofof parasitic
parasitic capacitance
capacitance is is large
large and
and cannot
cannot be
be ignored.
ignored. The
The irregularity
irregularity
rapidly changes, the influence of parasitic capacitance is large and cannot be ignored. The irregularity
and randomness
and randomness
randomness of of the
of the fractal
the fractal capacitor
fractal capacitor electrode
capacitor electrode will
electrode will help
will help
help toto reduce
to reduce the
reduce the influence
the influence
influence of of electric
electric field
of electric field
and field
strength
strength and
and parasitic
parasitic resonance
resonance and
and improve
improve the
the stability
stability of
of capacitive
capacitive response
response [22].
[22].
strength and parasitic resonance and improve the stability of capacitive response [22].
3. Methods
Methods
3. Methods
Fractal Electrode
3.1. Fractal Electrode Manufacture
3.1. Fractal Electrode Manufacture
For comparison,
comparison,both bothaatraditional
traditionalinterdigitated
interdigitated electrode
electrode gasgas sensor
sensor andand a fractal
a fractal electrode
electrode gas
For comparison, both a traditional interdigitated electrode gas sensor and a fractal electrode gas
gas sensor
sensor werewere designed
designed simultaneously.
simultaneously. SinceSince conventional
conventional processing
processing methods
methods cannot
cannot satisfy
satisfy the
sensor were designed simultaneously. Since conventional processing methods cannot satisfy the
the requirements
requirements of finger
of finger pitch
pitch andand finger
finger width
width of of electrodesasasrequired
electrodes requiredherehere(in(in the
the order
order of
requirements of finger pitch and finger width of electrodes as required here (in the order of
micrometers), microelectromechanical
micrometers), microelectromechanical system system (MEMS)
(MEMS) processing
processing was was used
used toto fabricate
fabricate the
the sensor
sensor
micrometers), microelectromechanical system (MEMS) processing was used to fabricate the sensor
copper clad
electrode. A copper clad plate
plate was
was selected
selected asas substrate,
substrate, which
which is
is not
not only
onlyinsulated
insulated from
from the
theoutside,
outside,
electrode. A copper clad plate was selected as substrate, which is not only insulated from the outside,
but also
also has
hasgood
goodadhesion
adhesiontotothe the electrodes
electrodes and and
was was convenient
convenient for dicing
for dicing in mass
in mass production.
production. The
but also has good adhesion to the electrodes and was convenient for dicing in mass production. The
The copper
copper layer,
layer, withwith a specified
a specified thickness,
thickness, was was
first first electroplated
electroplated ontoonto the substrate,
the substrate, then then
etchedetched
into
copper layer, with a specified thickness, was first electroplated onto the substrate, then etched into
intoelectrode
the the electrode structure
structure by abychemical
a chemical reaction
reaction method.
method. Finally,
Finally, a athin
thinlayer
layerofofgold
gold was
was deposited
the electrode structure by a chemical reaction method. Finally, a thin layer of gold was deposited
onto the surface of the electrode by a gold deposition process (pre-treatment, nickel deposition, gold
onto the surface of the electrode by a gold deposition process (pre-treatment, nickel deposition, gold
deposition and post-treatment) to increase conductivity and prevent oxidation. The structure of each
deposition and post-treatment) to increase conductivity and prevent oxidation. The structure of each
layer of the sensor is shown in FigureFigure 5 [23].
[23]. The
Theaspect
aspectratio
ratioand
andeffective
effective areas
areas of
of the
the fractal
fractal electrode
electrode
layer of the sensor is shown in Figure 5 [23]. The aspect ratio and effective areas of the fractal electrode
and thetheinterdigital electrode
interdigital was 1:1. The
electrode waselectrode
1:1. structure is shown instructure
The electrode Figure 6, wherein
is shown the widthin
and the interdigital electrode was 1:1. The electrode structure is shown in
of the plate
Figure is 400 µm,
6, wherein the pitch
the width is 200
of the µm,
plate and μm,
is 400 the thickness 30 µm.
the pitch is 200 μm, and the thickness is 30 μm.
Figure 6, wherein the width of the plate is 400 μm, the pitch is 200 μm, and the thickness is 30 μm.

(a) (b)
(a) (b)
Figure 5. Illustration of sensor structure. (a) Top view; (b) side view.
Figure 5. Illustration of sensor structure. (a) Top view; (b) side view.
Figure 5. Illustration of sensor structure. (a) Top view; (b) side view.
Chemosensors
Chemosensors 7, 0031
2019,
2019, 6, x 5 of
5 11
of 10

(a) (b)
Figure 6. Structure diagram of sensor plate. (a) Fractal electrode; (b) ordinary Euclidean electrode
(interdigital
Figure 6. electrode).
Structure diagram of sensor plate. (a) Fractal electrode; (b) ordinary Euclidean electrode
(interdigital electrode).
3.2. Carbon Nano Tube Coatings
3.2.
ToCarbon Nano the
investigate Tube Coatings
novel fractal electrode structure, we selected carboxylated multi-walled carbon
nanotubes, produced by Chengdu
To investigate the novel fractal Organic Chemicals
electrode Co., Ltd.
structure, we (Chengdu, China China),
selected carboxylated Chinese
multi-walled
Academy
carbon of Sciences, as
nanotubes, the basicby
produced building
Chengdumaterial. ThisChemicals
Organic was selected
Co.,asLtd.
our group were experienced
(Chengdu, China China),
with
Chinese Academy of Sciences, as the basic building material. This was selected through
depositing such films. It was prepared by catalytic cracking of natural gas a nickel
as our group were
catalyst to obtain multi-walled carbon nanotubes, which were then prepared by oxidation
experienced with depositing such films. It was prepared by catalytic cracking of natural gas of KM n O4
through
in H 2 SO4 solution.
a nickel The
catalyst to mainmulti-walled
obtain performancecarbon
parameters are shown
nanotubes, in were
which Tablethen
1 [24].
prepared by oxidation of
KMnO4 in H2SO4 solution. The main performance parameters are shown in Table 1 [24].
Table 1. Performance parameters of carboxylated multi-walled carbon nanotubes.
Table 1. Performance parameters
Characteristic Unit of MWCNTs
carboxylated multi-walled carbon
Characterization nanotubes.
Method
Characteristic
Outer Diameter Unitnm MWCNTs
20–30 Characterization
HRTEM (high resolutionmethod
Outer Diameter nm 20–30 transmission electron transmission
HRTEM (high resolution
microscope), Raman
electron microscope), Raman
PurityPurity wt%wt% >98 >98 TGA(thermal
TGA (thermal gravimetric
gravimetric analyzer) &
analyzer) & TEM
TEM(transmission
(transmission electron microscope)
electron
Length μm 10–30 TEM
microscope)
Special Surface Area
Length m2/gµm >11010–30 BET (Brunauer,
TEMEmmett, Teller) Specific
Surface Area Detection Method
Special Surface Area m2 /g >110 BET (Brunauer, Emmett,
ASH wt% <1.5 TGAArea
Teller) Specific Surface
Electric Conductivity s/cm >100 Detection Method -
Tap Density
ASH g/cmwt%
3 0.28<1.5 TGA -
-COOH Content wt% 1.23 XPS (X-ray photoelectron spectroscopy) &
Electric Conductivity s/cm >100 -
Titration
Tap Density g/cm3 0.28 -
We adopted-COOH Content
a combination of a wt%
drop coating1.23
and a spinXPS (X-raymethod
coating photoelectron
to prepare a translucent
spectroscopy) & Titration
conductive carbon nanotube film under optimized process parameters. CNTs were first dispersed is
a combination of a dispersant (TNWDIS, Carbon nanotube water dispersant, Chengdu Organic
Chemicals
We adoptedCo.,aLtd., Chinese Academy
combination of Sciences)
of a drop coating and aand
spindeionized water, which
coating method was afollowed
to prepare by an
translucent
ultrasoniccarbon
conductive treatment. Thisfilm
nanotube wasunder
proceeded by aprocess
optimized centrifuged treatment
parameters. CNTs towere
remove
first undispersed
dispersed
is aparticles. The of
combination electrode plates(TNWDIS,
a dispersant were cleaned
Carbonin nanotube
ethanol and washer
water in deionized
dispersant, Chengdu water before
Organic
deposition.
Chemicals Co.,Then
Ltd.,aChinese
small amount
Academy of dispersed CNTs
of Sciences) andwere dropped
deionized ontowhich
water, the electrodes whilst
was followed byitan
was
spun totreatment.
ultrasonic form a control layer
This was thickness.
proceeded byThe resistancetreatment
a centrifuged of the film was continuously
to remove undispersed measured
particles. to
Theensure eachplates
electrode coating
werewas of a similar
cleaned thickness.
in ethanol and washer in deionized water before deposition. Then a
small amount of dispersed CNTs were dropped onto the electrodes whilst it was spun to form a control
3.3.thickness.
layer Sensor Measurement
The resistance of the film was continuously measured to ensure each coating was of a
similar Tothickness.
measure the response of the gas sensor, it was connected in series with fixed resistor, as
shown in Figure 7. Here an AC (alternating current) sine wave signal generated by VirtualBench
virtual instrument (National Instrument, U.S.A) was used as the input signal and the voltage V1
across the sensor was detected by data acquisition card (DAQ, ART Technology, China). The
 Chemosensors 2019, 7, 0031 6 of 11

3.3. Sensor Measurement

Chemosensors 2019, 6, x 6 of 10
To measure the response of the gas sensor, it was connected in series with fixed resistor, as shown
Chemosensors
in Figure2019, 6, x an AC (alternating current) sine wave signal generated by VirtualBench6virtual
7. Here of 10
impedance characteristics of the sensor was proportional to the input the voltage V1 sensor. Thus, if
instrument (National Instrument, USA) was used as the input signal and the voltage V 1 across
we definecharacteristics
impedance the impedance ofthe
sensor to be Z,proportional
and the voltage of the signal
the source be
V1U, we have:
the sensor was detected of by datasensor was
acquisition card (DAQ,toARTthe input voltage
Technology, China).sensor. Thus, if
The impedance
we define the impedance of sensor to be Z, and
characteristics of the sensor was proportional the voltage
→ to the Z of the signal source be U, we have:
input→the voltage V 1 sensor. Thus, if we define the
impedance of sensor to be Z, and the voltage V =
→ of the →U source be U, we have: (1)
Z signal
V1 = R1 + ZU
1
(1)
→ R +Z Z →
1
V1 = U (1)
R1 + Z
V1
V1

C
C R1
R R1
R AC
AC
U
U

Figure 7. Experimental circuit.

Figure 7. Experimental circuit.
After absorbing
absorbing gases,
gases,the theimpedance
impedanceofofthe thesensor
sensor changes,
changes, wherein
whereinthethe
voltage change
voltage changeacross the
across
sensor,
the After inabsorbing
sensor, theincircuit, was measured
the circuit,
gases, was
the and corresponding
measured
impedance ofand gas concentration
corresponding
the sensor changes, gas was obtained
concentration
wherein the voltage was byobtained
changecalculation.
acrossby
Under test,
calculation. the
Undersensors
test,were
the exposed
sensors to 3
were min of
exposed drytoclean
3 minair,
offollowed
dry clean
the sensor, in the circuit, was measured and corresponding gas concentration was obtained by by an
air, injection
followed of
bythe
an target gas
injection
(nitric oxide
of the target
calculation. in this case)
gas (nitric
Under for
oxide
test, the a further 3
in thiswere
sensors min. Different
case)exposed
for a further concentrations
3 min.
to 3 min of Different of NO gas
dry cleanconcentrationswere
air, followed by used
ofan in the
NOinjection tests
gas were
from
used 2
into 10
the ppm.
tests The
from air
2 chamber
to 10 ppm. was
Thea 325
air ml stainless
chamber was steel
a box
325 mlcoated with
stainless
of the target gas (nitric oxide in this case) for a further 3 min. Different concentrations of NO gas were Teflon.
steel box Sensors
coated were
with
welded
Teflon. on a
Sensors printed
were circuit
welded board
on a (PCB) and
printed placed
circuit upside
board down
(PCB) andin the box.
placed
used in the tests from 2 to 10 ppm. The air chamber was a 325 ml stainless steel box coated with All
upside sensors
down operated
in the at
box.
room
Teflon. temperature.
All sensors
Sensors operated A
were weldedmass
at room flow controller
ontemperature. (MFC)
A mass
a printed circuit was
board used
flow to control
controller
(PCB) gas
(MFC)
and placed flow rate.
was used
upside The test system
downtoincontrol
the box. gasis
shown
flow
All sensorsin Figure
rate. The test8.system
operated at room is shown in Figure
temperature. 8.
A mass flow controller (MFC) was used to control gas
flow rate. The test system is shown in Figure 8.

Figure 8. Gas sensing test system.

Figure 8. Gas sensing test system.
The sensor
The sensor response
response is
is defined
defined as:
as:

dVdV==((VV−−V
V0 ))/V
The sensor response is defined as: V0 (2)
0 0 (2)
dV = (V − V0 ) V0 (2)
where V represent sensor response at current concentration, V0 represent sensor response at 0 ppm
where V represent sensor response at current concentration, V 0 represent sensor response at 0 ppm
of stimulus.
where V represent sensor response at current concentration, V0 represent sensor response at 0 ppm
of stimulus.
of stimulus.
4. Results and Discussion
4. Results and Discussion
4.1. Morphology
4.1. Morphology
The SEM (scanning electron microscopy) image of the MWNTs (Multi-walled carbon nanotubes)
filmThe
obtained is shownelectron
SEM (scanning in Figure 9. Image at image
microscopy) the topofleft
theisMWNTs
enlarged(Multi-walled
to a scale of 500 μm, which
carbon shows
nanotubes)
Chemosensors 2019, 7, 0031 7 of 11

4. Results and Discussion

4.1. Morphology
The SEM
Chemosensors 2019,(scanning
6, x electron microscopy) image of the MWNTs (Multi-walled carbon nanotubes) 7 of 10
film obtained is shown in Figure 9. Image at the top left is enlarged to a scale of 500 µm, which shows
that
that the carbon
carbonnanotubes
nanotubeswerewereuniformly
uniformly deposited
deposited in gaps
in gaps between
between the fractal
the fractal electrodes.
electrodes. The
The image
image at thewas
at the right right was enlarged
enlarged to aofscale
to a scale of 20
20 µm, μm,which
from from awhich a stacked
stacked patternpattern of multi-walled
of multi-walled carbon
carbon nanotubes
nanotubes and between
and spaces spaces between
nanotubesnanotubes can be
can be seen. Theseen. The detected
detected gas wasof
gas was capable capable of being
being adsorbed
adsorbed in these
in these spaces, spaces,in
resulting resulting
physicalin
orphysical
chemicalorchanges,
chemical changes,
thereby thereby
changing thechanging
electricalthe electrical
conductivity
conductivity or dielectric
or dielectric constant constant
of the carbonof the carbon
nanotubes asnanotubes
a medium.as a medium.

Figure
Figure 9.9. SEM
SEMimages
imagesofofmulti-walled
multi-walledcarbon
carbonnanotube
nanotubefilms
filmsatatdifferent
differentmagnifications
magnifications(scale
(scalebars,
bars,
left:
left: 11mm;
mm;right:
right:20
20μm).
µm).

Inorder
In ordertotofurther
furtherdetect
detectthe
thethickness
thicknessbetween
betweenthe thesensor
sensorelectrode
electrodeand
andthethefilm,
film,aaVeeco
VeecoDekak
Dekak
150 surface
150 surface probe
probe profilometer
profilometer manufactured
manufactured by byBruker
BrukerVeeco
Veeco was
was used
used to
to test
test the
the surface
surface thickness
thickness
and roughness
and roughness ofof the
the sensor
sensor film.
film. The
Thetest
testresult
result indicated
indicated that
that the
the deposited
deposited MWCNT
MWCNTfilm film had
had aa
thickness of
thickness of between
between 24–27
24–27 μm.
µm.

4.2. Gas
4.2. GasSensing
SensingCharacteristic
Characteristic
In order
In order to
to verify
verify the
the repeatability
repeatability of of the
the sensor,
sensor, we
we manufactured
manufactured two two sensors
sensors with
with fractal
fractal
electrodes and three with interdigital electrodes. To test the basic functionality of
electrodes and three with interdigital electrodes. To test the basic functionality of the sensor (beforethe sensor (before
AC measurements),
AC measurements),we wemeasured
measured thethe
resistance
resistanceacross the sensor
across using ausing
the sensor simpleamultimeter (VICTOR,
simple multimeter
China) andChina)
(VICTOR, exposedandthe sensor the
exposed to 2–10
sensorppm of NOppm
to 2–10 in 2of
ppm
NOsteps. Thesteps.
in 2 ppm obtained
Theresistances are shown
obtained resistances
in Figure 10. As can be seen the resistance of all the sensors changed in a similar
are shown in Figure 10. As can be seen the resistance of all the sensors changed in a similar way, way, as the gas
as
concentration increased. However, their initial resistance values (resistance of sensor
the gas concentration increased. However, their initial resistance values (resistance of sensor under under clean air)
were air)
clean different, which is due
were different, to the
which discrepancy
is due of the MWCNTs
to the discrepancy film in thefilm
of the MWCNTs coating
in theprocess.
coating process.
electrodes and three with interdigital electrodes. To test the basic functionality of the sensor (before
AC measurements), we measured the resistance across the sensor using a simple multimeter
(VICTOR, China) and exposed the sensor to 2–10 ppm of NO in 2 ppm steps. The obtained resistances
are shown in Figure 10. As can be seen the resistance of all the sensors changed in a similar way, as
the gas concentration
Chemosensors 2019, 7, 0031 increased. However, their initial resistance values (resistance of sensor under
8 of 11
clean air) were different, which is due to the discrepancy of the MWCNTs film in the coating process.

Chemosensors 2019, 6, x Figure 10. Resistance versus concentration curves of the five sensors. 8 of 10

The changein
The change insensitivity
sensitivityofof the
the sensor
sensor with
with NONO concentration
concentration is shown
is shown in Figure
in Figure 11 below.
11 below. Here
Here we see the gas response following a traditional power law
we see the gas response following a traditional power law model. model.

Figure 11. Curve

Curve of
of sensor
sensor responses to NO with standard error bars at concentration gradient.

From the initial DC measurements, we were able to confirm that our sensors responded to NO
across the required concentration range. Thus,
Thus, we
we then considered
considered the impedance
impedance of the sensors at
four different frequencies (50,
(50, 100,
100, 250 and 500
500 kHz).
kHz). These were chosen to give a broad spectrum of
different
different impendence
impendencevalues.
values.The sensors
The were
sensors then
were tested
then to the
tested tosame NO concentrations
the same used earlier.
NO concentrations used
The results
earlier. Theof theseof
results measurements are shown
these measurements arein Figurein12.
shown Figure 12.

(a) (b)
 From the initial DC measurements, we were able to confirm that our sensors responded to NO
across the required concentration range. Thus, we then considered the impedance of the sensors at
four different frequencies (50, 100, 250 and 500 kHz). These were chosen to give a broad spectrum of
different impendence
Chemosensors 2019, 7, 0031 values. The sensors were then tested to the same NO concentrations 9used
of 11
earlier. The results of these measurements are shown in Figure 12.

(a) (b)

(c) (d)
Figure2019,
Chemosensors Comparison
12. 6, x of sensor responses between the two structures. (a) Response at 50 kHz;
9 of 10
Figure
(b) 12. Comparison
response of response
at 100 kHz; (c) sensor responses
at 250 kHz;between the two
(d) response structures.
at 500 kHz. (a) Response at 50kHz;
(b) response
As it can beatseen
100 kHz;
from(c)Figure
response
12,atthe
250 response
kHz; (d) response at 500ofkHz.
amplitude the fractal electrode sensor was
As than
greater it canof
bethe
seen from Figure
interdigital 12, the response
electrode amplitude
sensor under of theconditions.
the same fractal electrode
Thus,sensor wasthat
it shows greater
the
than of the interdigital electrode sensor under the same conditions. Thus, it shows
fractal electrode gas sensor outperformed, in terms of sensitivity, those formed from traditional that the fractal
electrode geometries.
electrode gas sensor outperformed,
Evaluating only in the
terms of sensitivity,
fractal those
sensors, we thenformed from the
considered traditional electrode
sensitivity and a
geometries. Evaluating only the fractal sensors, we then considered the sensitivity
function of frequency over the NO gas range defined earlier. The results of this are shown in and a function of
frequency
Figure over the NO gas range defined earlier. The results of this are shown in Figure 13 below.
13 below.

Response of
Figure 13. Response of the
the fractal
fractal sensor
sensor to
to NO
NO gases
gases at different frequencies.

This figure shows that the higher the NO concentration, the greater
greater the
the difference of sensor
response at different frequencies. Moreover, the response amplitude was large
response at different frequencies. Moreover, the response amplitude was large at at low
low frequencies,
frequencies,
while the linearity of the response curve was not so good. These characteristics need more
investigation in subsequent experiments. Depending on the concentration of the gas being measured,
the frequency of signal source can be adjusted to optimize the response of sensor, which helps to
extend the range of concentration detection. At the same time, for different types of gases to be tested,
adjusting the sensor response at different frequencies may improve features among different types
Chemosensors 2019, 7, 0031 10 of 11

while the linearity of the response curve was not so good. These characteristics need more investigation
in subsequent experiments. Depending on the concentration of the gas being measured, the frequency
of signal source can be adjusted to optimize the response of sensor, which helps to extend the range of
concentration detection. At the same time, for different types of gases to be tested, adjusting the sensor
response at different frequencies may improve features among different types of gases, which helps the
sensor to distinguish gas species.

5. Conclusions
In this paper, the influence of the shape structure of a sensor on its gas-sensitive characteristics
was studied. The fractal theory was introduced into the structure design of resistance and capacitance
(RC)-based gas sensors. The structure of the film on the surface of the sensor was examined by physical
methods such as scanning electron microscopy and a surface probe profilometer. With NO as test gas,
the effectiveness of the sensor was proved by experiments. It is showed that the design method can
effectively improve the sensitivity of gas sensor and provide a new method for the development of
high performance gas sensors.
In our experiments, the sensors showed poor linearity. In the future, nanomaterials such as
graphene or high molecular polymer may be considered as replacements. At the same time, electrodes
of the sensor can be further miniaturized by means of a microelectromechanical system. Finally, in the
future, the selectivity of the sensor can be further verified by detecting multiple gases.

Author Contributions: Conceptualization, F.T. and J.A.C.; data curation, T.Y., F.X.; formal analysis, F.T., T.Y.;
funding acquisition, J.A.C., R.L.; investigation, A.J., Y.H.; methodology, T.Y., F.T., Y.X., J.Q.; project administration,
F.T.; resources, Y.X., J.Q., R.L.; software, Y.H., T.Y.; supervision, F.T.; validation, F.X., R.L.; visualization, Z.W., R.L.,
J.Q.; writing–original draft preparation, T.Y.; writing—review and editing, J.A.C., Z.W.
Funding: This research was funded by the Common Crucial Technique Innovation Project of Chongqing
Key Industries (cstc 2016zdcy-ztzx007-03), the National International Science and Technology Cooperation
Project (2014DFA31560), and the Chongqing Research Program of Basic Research and Frontier Technology (No.
cstc2016jcyjA0103).
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Li, S.; Wang, W.; Liang, F.; Zhang, W.X. Heavy metal removal using nanoscale zero-valent iron (nZVI): Theory
and application. J. Hazard. Mater. 2016, 322, 163–171. [CrossRef] [PubMed]
2. Chen, X. Method for Improving Detection Accuracy of Electronic Nose System. Master’s Thesis,
Chongqing University, Chongqing, China, 2016.
3. Zhang, W. Research on Key Technologies of Gas Identification Based on Sensor Array Transient Response.
Ph.D. Thesis, National University of Defense Technology, Changsha, China, 2013.
4. Schroeder, V.; Savagatrup, S.; He, M.; Lin, S.; Swager, T.M. Carbon Nanotube Chemical Sensors. Chem. Rev.
2019, 119, 1599–1663. [CrossRef] [PubMed]
5. Ishihara, S.; Labuta, J.; Nakanishi, T.; Tanaka, T.; Kataura, H. Amperometric Detection of Sub-ppm
Formaldehyde Using Single-Walled Carbon Nanotubes and Hydroxylamines: A Referenced Chemiresistive
System. ACS Sens. 2017, 2, 1405–1409. [CrossRef] [PubMed]
6. Wang, X. Fabrication of Porous Metal Oxide Semiconductors and Analysis of Their Properties. Ph.D. Thesis,
Shandong University, Jinan, China, 2014.
7. Gillet, M.; Aguir, K.; Bendahan, M.; Mennini, P. Grain size effect in sputtered tungsten trioxide thin films on
the sensitivity to ozone. Thin Solid Films 2005, 484, 358–363. [CrossRef]
8. Xiao, C. Study on the One-Dimensional Structure Design of Nanomaterials and Their Gas Sensing Properties.
Master’s. Thesis, Jilin University, Changchun, China, 2015.
9. Bo, L. Study on Structure Improvement of Microstructured Si-based SnO2 Gas Sensor. Master’s Thesis,
Dalian University of Technology, Dalian, China, 2010.
10. Zhang, Q.; Guan, Z.-S. Gas Sensors of Electric Resistance Semiconductors. Instrum. Tech. Sens. 2006, 7, 6–9.
Chemosensors 2019, 7, 0031 11 of 11

11. Ma, X. Study on Design and Characteristics of Micro-Structure Gas Sensor Based on Silicon Substrate.
Master’s Thesis, Jilin University, Changchun, China, 2013.
12. Lin, X. Philosophical Random Walk in Fractals; Capital Normal University Press: Beijing, China, 1999.
13. Iijima, S. Helical microtubules of graphitic carbon. Nature 1991, 354, 56–58. [CrossRef]
14. Zhou, C.; Kong, J.; Dai, H. Electrical measurements of individual semiconducting single-walled carbon
nanotubes of various diameters. Appl. Phys. Lett. 2000, 76, 1597–1599. [CrossRef]
15. Hannon, A.; Lu, Y.; Li, J.; Meyyappan, M. Room temperature carbon nanotube based sensor for carbon
monoxide detection. J. Sens. Sens. Syst. 2014, 3, 349–354. [CrossRef]
16. Kong, J.; Franklin, N.R.; Zhou, C.; Chapline, M.G.; Peng, S.; Cho, K.; Dai, H. Nanotube molecular wires as
chemical sensors. Science 2000, 287, 622–625. [CrossRef] [PubMed]
17. Li, J.; Lu, Y.; Ye, Q.; Cinke, M.; Han, J.; Meyyappan, M. Carbon Nanotube Sensors for Gas and Organic Vapor
Detection. Nano Lett. 2003, 3, 929–933. [CrossRef]
18. Hoa, N.D.; Quy, N.V.; Cho, Y.; Kim, D. An ammonia gas sensor based on non-catalytically synthesized carbon
nanotubes on an anodic aluminum oxide template. Sens. Actuators B Chem. 2007, 127, 447–454. [CrossRef]
19. Peano, G. Sur une courbe, qui remplit toute une aire plane. Math. Ann. 1970, 36, 157–160. [CrossRef]
20. Hilbert, D. Über die stetige Abbildung einer Linie auf ein Flächenstück. Math. Ann. 1891, 38, 459–460.
[CrossRef]
21. Frame, M.; Urry, A. Fractal Worlds: Grown, Built, and Imagined; Yale University Press: New Haven, CT,
USA, 2016.
22. Xu, W.; Zhuang, M.; Wang, D. Fractal Planar Capacitors Based on Interdigital Structures. J. Jimei Univ.
Nat. Sci. 2010, 15, 389–393.
23. Deng, H. Research of the MWNTs Benzene Series Vapor Sensor. Master’s Thesis, Harbin Institute of
Technology, Harbin, China, 2010.
24. Chengdu Organic Chemicals Co., Ltd. Chinese Academy of Sciences. Multi-Walled Carbon Nanotube
Material Technical Data [EB/OL]. Available online: http://meizhenqu.l58a.bolead.com/view.php?prt=3,29,50,
81&id=114 (accessed on 6 May 2019).

© 2019 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/).