coatings

Article
A Fully Inkjet-Printed Strain Sensor Based on
Carbon Nanotubes
Hsuan-Ling Kao 1,2,3, * , Cheng-Lin Cho 1 , Li-Chun Chang 4,5 , Chun-Bing Chen 6 ,
Wen-Hung Chung 6 and Yun-Chen Tsai 1
1 Department of Electronic Engineering, Chang Gung University, Tao-Yuan 33302, Taiwan;
iptclc@gmail.com (C.-L.C.); np260369@gmail.com (Y.-C.T.)
2 Centre for Reliability Sciences and Technologies, Chang Gung University, Tao-Yuan 33302, Taiwan
3 Department of Dermatology, Chang Gung Memorial Hospital, Linkou Branches, Tao-Yuan 33305, Taiwan
4 Department of Materials Engineering, Ming Chi University of Technology, New Taipei 243303, Taiwan;
lcchang@mail.mcut.edu.tw
5 Center for Thin Film Technologies and Applications, Ming Chi University of Technology,
New Taipei 243303, Taiwan
6 Department of Dermatology, Chang Gung Memorial Hospital, Keelung, Taipei, and Linkou Branches,
Tao-Yuan 33305, Taiwan; chunbing.chen@gmail.com (C.-B.C.); wenhungchung@yahoo.com (W.-H.C.)
* Correspondence: snoopy@mail.cgu.edu.tw; Tel.: +886-3-211-8800 (ext. 5951)




Received: 22 July 2020; Accepted: 13 August 2020; Published: 14 August 2020


Abstract: A fully inkjet-printed strain sensor based on carbon nanotubes (CNTs) was fabricated
in this study for microstrain and microcrack detection. Carbon nanotubes and silver films were
used as the sensing layer and conductive layer, respectively. Inkjet-printed CNTs easily undergo
agglomeration due to van der Waals forces between CNTs, resulting in uneven films. The uniformity
of CNT film affects the electrical and mechanical properties. Multi-pass printing and pattern
rotation provided precise quantities of sensing materials, enabling the realization of uniform CNT
films and stable resistance. Three strain sensors printed eight-layer CNT film by unidirectional
printing, rotated by 180◦ and 90◦ were compared. The low density on one side of eight-layer CNT
film by unidirectional printing results in more disconnection and poor connectivity with the silver
film, thereby, significantly increasing the resistance. For 180◦ rotation eight-layer strain sensors,
lower sensitivity and smaller measured range were found because strain was applied to the uneven
CNT film resulting in non-uniform strain distribution. Lower resistance and better strain sensitivity
was obtained for eight-layer strain sensor with 90◦ rotation because of uniform film. Given the uniform
surface morphology and saturated sheet resistance of the 20-layer CNT film, the strain performance
of the 20-layer CNT strain sensor was also examined. Excluding the permanent destruction of the first
strain, 0.76% and 1.05% responses were obtained for the 8- and 20-layer strain sensors under strain
between 0% and 3128 µε, respectively, which demonstrates the high reproducibility and recoverability
of the sensor. The gauge factor (GF) of 20-layer strain sensor was found to be 2.77 under strain
from 71 to 3128 µε, which is higher than eight-layer strain sensor (GF = 1.93) due to the uniform
surface morphology and stable resistance. The strain sensors exhibited a highly linear and reversible
behavior under strain of 71 to 3128 µε, so that the microstrain level could be clearly distinguished.
The technology of the fully inkjet-printed CNT-based microstrain sensor provides high reproducibility,
stability, and rapid hardness detection.

Keywords: inkjet printing; carbon nanotubes; strain sensors

Coatings 2020, 10, 792; doi:10.3390/coatings10080792 www.mdpi.com/journal/coatings

Coatings 2020, 10, 792 2 of 15

1. Introduction
Strain sensors have great potential for application in wearable electronic devices and smart
skin to monitor human motion, personal health monitoring, and human–machine interfaces. Strain
sensors convert mechanical deformation into electrical signals such as capacitance or resistance to
monitor changes in strain, force, pressure, and shape. Carbon nanotubes (CNTs) possess impressive
properties, such as a hundred times higher tensile strength than steel, low electrical conductivity,
and good thermal conductivity [1,2]. Many researchers have investigated CNTs as pressure sensors [3,4]
and strain sensors [5,6] because of their excellent electromechanical properties. Bu et al. reported
drop-casting-prepared multi-walled CNT films with a sensitivity of 2.5 for the strain below 0.1% [7].
Lee et al. proposed a spray-coated single-walled CNT film with good linearity over a microstrain
range of 0 to 400 µε [8]. Sahatiya et al. deposited multi-walled CNTs on an eraser by rod-coating
and obtained a strain-and-pressure sensor with a strain gauge factor (GF) of 2.4 and capacitive
pressure sensor sensitivity of 0.135 MPa−1 [9]. Chun et al. reported highly oriented and free-standing
hydrophobic CNT sheets synthesized by chemical vapor deposition and transferred onto polyethylene
naphthalate substrate; the CNT sheets exhibited sensitive piezoresistive responses to applied pressures
of 0.1–40 kPa [10]. Previous studies have presented the fabrication technologies of CNT films such
as drop-casting, spray coating, rod-coating, and chemical vapor deposition, which are complex and
laborious processes results in high manufacturing costs and generate large material waste. In addition,
effectively controlling the precise amount of sensing materials using the above processes is difficult.
Inkjet printing technology is a fast, convenient, low-fabrication-temperature, and low-cost
technique. The ability of inkjet printing is to directly print patterns layer-by-layer based on digital
images to provide additive manufacturing, without involving any photolithography process [11–14].
Moreover, inkjet-printed technology can be claimed to be environmentally friendly due to the savings
in materials and the reduction in the amount of waste generated which provides more economical than
other additive manufacturing technologies [15–17]. Filling-in an A4 size paper only requires 0.75 mL of
silver ink, about U.S. $7.5. The filling-in an A4 paper is about 2 h with a frequency of 20 kHz. Carbon
nanotubes are common materials used for printing on various substrates based to obtain accurate
and controllable conductive circuits and sensors [18–21]. A counterfeit fingerprint with electrical and
geometric patterns of exquisite ridges and furrows of one to five layers was printed on paper using
an inkjet printer; the printed patterned CNT film electrode as the top plate of the capacitive sensor
could unlock a smartphone [22]. Some researchers in a previous study fabricated an inkjet-printed
20-layer CNT-based strain sensor with a 4.7% standard deviation over 20 cycles for strains between 0
and 800 µε [23]. An inkjet-printed carbon nanotube embedded thin film is able to capture non-uniform
strain distribution at 0.001% strain level [24]. Therefore, the inkjet printing technology can be precisely
controlled the sensing material based on the ink content or number of layers. However, CNT ink
is difficult to disperse CNTs due to van der Waals forces between CNTs. The efficiency of CNT
dispersion in the ink affects the uniformity of CNT films and even cause nozzle clogging. Inkjet-printed
CNT film is easy to agglomerate. Agglomerates CNT films deteriorate the electrical and mechanical
properties because of microscopic inhomogeneity. CNTs can agglomerate at the periphery during
inkjet printing (so-called the coffee ring effect), resulting in an uneven film and unstable resistance [25].
Added to the surfactant in the ink such as sodium dodecyl benzene sulfonate to improve continuity
and high uniformity of the inkjet-printed CNT films was presented [26]. However, the rinsing process
performed every two layers to remove most of the residual surfactant and solvent is unavoidable and
time-consuming. In the current work, an aqueous conductive ink of multi-walled carbon nanotubes
prepared by ball-milling dispersion method was used to provide good dispersion. Multi-pass-printing
and pattern rotation were used to improve the coffee ring effect and obtain stable resistance and a
uniform CNT film. The morphologies and electrical characteristics of an inkjet-printed CNT film as a
strain-sensing layer were studied. Two varieties of ink, namely silver nanoparticle ink and CNT ink,
were effectively arranged on a substrate according to digital images. A fully inkjet-printed CNT-based
microsensor was fabricated using CNTs and silver inks. The electrical and mechanical properties of
Coatings 2020, 10, 792 3 of 15
Coatings 2020, 10, x FOR PEER REVIEW 3 of 16

three ◦ and 90◦

layer strain sensors
CNT film printed eight-layer
by unidirectional CNTrotated
printing, film by byunidirectional
180° and 90°printing, rotated byThe
were compared. 180CNT film
were compared. The CNT film printed with 90 ◦ rotation provides lower resistance and better strain
printed with 90° rotation provides lower resistance and better strain sensitivity for electrical and
sensitivity
mechanicalfor electricalThe
properties. andGFmechanical
of 20-layerproperties.
strain sensorThewas
GF 2.77
of 20-layer strainfrom
under strain sensor71was 2.77µε.
to 3128 under
The
strain from 71 to 3128 µε. The microstrain sensor electromechanical properties
microstrain sensor electromechanical properties were characterized to validate its feasibility,were characterized
to validate its and
repeatability, feasibility, repeatability,
reversibility. and inkjet-printed
The fully reversibility. The fully
strain inkjet-printed
sensor strain sensor
with microstrain with
detection
microstrain
capability candetection
be usedcapability can bescleroderma
to distinguish used to distinguish
hardnessscleroderma hardness cracks.
and detect thin-film and detect
Thethin-film
article is
cracks. The article is organized as follows: Section 2 outlines the fabrication process
organized as follows: Section 2 outlines the fabrication process and experiment procedure. Section and experiment
procedure.
3.1 presentsSection 3.1 presents the
the characteristics characteristicsCNTs
of inkjet-printed of inkjet-printed
by multi-pass CNTs by multi-pass
printing printing
and pattern and
rotation.
pattern rotation.
Section 3.2 focusesSection
on the 3.2 focuses
electrical on the electrical
properties of strainproperties of strain
sensors using bulgesensors usingconcluding
test. Finally, bulge test.
Finally,
remarksconcluding
are offered.remarks are offered.

2. Experiments
2. Experiments
In
In this
thisstudy,
study, a 100
a 100µmµmCLTE-MW
CLTE-MW (Rogers Corp.,Corp.,
(Rogers Chandler, AZ, USA)
Chandler, AZ,wasUSA)selected
was as the substrate.
selected as the
CLTE-MW substrate are ceramic filled, woven glass reinforced Polytetrafluoroethene
substrate. CLTE-MW substrate are ceramic filled, woven glass reinforced Polytetrafluoroethene composites
according
compositestoaccording
the manufacturer datasheetsdatasheets
to the manufacturer they exhibit theygood
exhibit mechanical
good mechanicalproperties for bulge
properties test,
for bulge
such as flexural strength of 113 MPa, tensile strength of 83 MPa, and
test, such as flexural strength of 113 MPa, tensile strength of 83 MPa, and flex modulus of 6463 flex modulus of 6463 MPa [27].
The
MPaelectrodes
[27]. Thewas pattern defined
electrodes using defined
was pattern silver nanoparticle
using silver inknanoparticle
(DGP-40LT-15C, ink Advanced
(DGP-40LT-15C,Nano
Products Co., Ltd., Sejong, Korea) and the sensing layer was fabricated
Advanced Nano Products Co., Ltd., Sejong, Korea) and the sensing layer was fabricated using carbon using carbon nanotube ink
(CNT-22, lab311, Ltd., Seoul, Korea). Silver nanoparticle ink contained
nanotube ink (CNT-22, lab311, Ltd., Seoul, Korea). Silver nanoparticle ink contained 30 to 35 wt.% 30 to 35 wt.% silver loading,
dispersed in triethylene
silver loading, dispersedglycol monoethyl
in triethylene ethermonoethyl
glycol with a viscosity
ether of 10–18
with cP and surface
a viscosity of 10–18 tension
cP and of
30–40 dyne/cm. The viscosity and surface tension of the CNT ink were
surface tension of 30–40 dyne/cm. The viscosity and surface tension of the CNT ink were measured measured as 2.5 ± 0.1 cP and
40
as mN/m,
2.5 ± 0.1respectively.
cP and 40 mN/m, The CNT and silverThe
respectively. films
CNT were andprinted usingwere
silver films a Dimatix
printed DMPusing 2831 printer
a Dimatix
(FUJIFILM,
DMP 2831 printerValhalla,(FUJIFILM,
NY, USA) Valhalla,
equippedNY, withUSA) a DMC-11610
equipped cartridge. The cartridge
with a DMC-11610 contained
cartridge. The
16 nozzles, 21 µm in diameter and each nozzle generated 10-pL
cartridge contained 16 nozzles, 21 µm in diameter and each nozzle generated 10-pL ink drops. ink drops. The substrate holder
The
comprises of a built-in heater ◦ C.
substrate holder comprises ofable to heat
a built-in the substrate
heater able to heat up to 60substrate
the up to 60 °C.
Figure
Figure 11 illustrates
illustrates thethefabrication
fabricationofofthe thefully
fullyinkjet-printed
inkjet-printed strain
strain sensor.
sensor. Before
Before printing,
printing, the
the CLTE-MW surface was treated via plasma polymerization to obtain
CLTE-MW surface was treated via plasma polymerization to obtain the necessary hydrophilicity for the necessary hydrophilicity
for printing.
printing. TheThe
plasmaplasma activation
activation was was performed
performed usingusing
oxygenoxygen
gas atgas at a pressure
a pressure of 9.33 ofPa
9.33
byPaanby an
radio
radio frequency (RF) generator operating at 13.56 MHz. Two copper
frequency (RF) generator operating at 13.56 MHz. Two copper films on CLTE were reserved as wire- films on CLTE were reserved as
wire-bonding pads for measurement. A five-layer silver film was printed
bonding pads for measurement. A five-layer silver film was printed onto the CLTE substrate using onto the CLTE substrate
using silver nanoparticle
silver nanoparticle ink on inka 60on 60 ◦ C thermal
°Cathermal chuck and chuckthenand thenonbaked
baked a 100 on
◦ C hard plate for
a 100plate
°C hard for 30 min to
30 min to
ensure ensureSubsequently,
drying. drying. Subsequently,
a multilayera multilayer
sensing sensing
film wasfilm was inkjet-printed
inkjet-printed using CNTusingink CNT at ink
room at
room temperature andbaked
then baked on aplate
hardatplate at for ◦
1501 h.C for 1 h. Multi-pass
temperature and then on a hard 150 °C Multi-pass printingprinting
and patternand rotation
pattern
rotation
was employed to improve the coffee ring effect and obtain a uniform CNT film. The CNT CNT
was employed to improve the coffee ring effect and obtain a uniform CNT film. The films
films printed by unidirectional and rotated by 180 ◦ and 90◦ were compared. Then, a second silver film
printed by unidirectional and rotated by 180° and 90° were compared. Then, a second silver film was
was printed by passes
five passes ◦ C thermal chuck, with 10 min delay per pass printing to avoid
printed by five on a on a 60thermal
60 °C chuck, with 10 min delay per pass printing to avoid silver
silver ink flowing into the film.
CNTFinally,
film. Finally, the were
films dried
were dried in a baking ◦ for 2 h.
ink flowing into the CNT the films in a baking oven oven
at 150at°C150for C
2 h.

Figure
Figure 1. The fabrication
1. The fabrication of
of the
the fully
fully inkjet-printed
inkjet-printed strain
strain sensor.
sensor.
Coatings 2020, 10, 792 4 of 15
Coatings 2020, 10, x FOR PEER REVIEW 4 of 16

Micro-Raman
Micro-Raman spectroscopy
spectroscopy with with excitation
excitation byby aa 532-nm
532-nm greengreen laser
laser atat aa current
current ofof 0.84
0.84 A was
A was
used
used to study the
to study the characteristic
characteristic features
features of of inkjet-printed
inkjet-printed CNTCNT film.film. Scanning
Scanning electron
electron microscopy
microscopy
(SEM, S3400N, Hitachi, Tokyo, Japan) under a 15-kV accelerating voltage
(SEM, S3400N, Hitachi, Tokyo, Japan) under a 15-kV accelerating voltage was employed to obtain was employed to obtain
the
the
surface morphologies and thicknesses of CNT films. Electrical resistance was measured using ausing
surface morphologies and thicknesses of CNT films. Electrical resistance was measured a
digital
digital multimeter
multimeter (Rigol DM3058E,
(Rigol DM3058E, Rigol Technologies,
Rigol Technologies, Inc., Beaverton,
Inc., Beaverton, NY, USA) NY,toUSA) to examine
examine the
the tensile
tensile strain by the bulge test method. Three samples with a length of 3 mm
strain by the bulge test method. Three samples with a length of 3 mm and a width of 1 mm for various and a width of 1 mm
for various
printing printing
passes werepasses were Sheet
measured. measured. Sheetvalue
resistance resistance
can bevalue can beby
calculated calculated by thevalue
the resistance resistance
from
value from number
number of squares. of squares.
Figure
Figure 2a 2ashows
showsthe the bulge
bulge testtest fixture
fixture withwith an o-ring
an o-ring of 24-mm
of 24-mm inner diameter
inner diameter to madeto themade
chamberthe
chamber airtight.toOwing
airtight. Owing to thebeing
the sample sample being face-up,
face-up, the bulge thetest
bulge test was
fixture fixture was placed
placed inside ainside
sealed a sealed
box to
box to isolate environmental interference. Nitrogen was passed through the
isolate environmental interference. Nitrogen was passed through the sealed box to maintain dryness. sealed box to maintain
dryness.
The pressureThe inpressure in bulge was
bulge chamber chamber was also controlled
also controlled by introducing
by introducing N2 throughNa2 pressure
through aregulating
pressure
regulating
valve. The valve.
sample The sample
bulges bulgeswhen
outward outward the when
pressurethecontrolled
pressure controlled
by nitrogen by gas
nitrogen gas isthan
is smaller smaller
the
than the cavity pressure. Two time controllers were used to switch pressure
cavity pressure. Two time controllers were used to switch pressure and maintain the pressure in a and maintain the
pressure
cavity forinthea cavity
requiredfor time.
the required
Strain (ε)time. Strain
is the (ε) is the between
relationship relationship the between
deformation the deformation
and the originaland
the original length of the CLTE substrate, which can be derived from the
length of the CLTE substrate, which can be derived from the geometry relationship in Figure 2b and geometry relationship in
Figure 2b and expressed
expressed as follows [28]: as follows [28]:
Rθ
𝑅θ − 1 = 2ℎ 2h2
εε =
= r − 1 = 3a2 (1)
(1)
𝑟 3𝑎
where aa is
where is the
the radius
radius of
of the o-ring, hh is
the o-ring, is the
the height
height of
of the
the substrate
substrate at
at the
the center, R is
center, R is the
the radius
radius ofof
curvature of
curvature of deforming
deforming substrate,
substrate, and
and θ θ is
is half
half the
the angle
angle subtended
subtended byby aa dome
dome surface
surface at
at its
its center
center of
of
curvature. The height was measured 10 times under various pressures using a height
curvature. The height was measured 10 times under various pressures using a height gauge, and the gauge, and the
average value
average value was
was taken.
taken.In
Inthis
thiswork,
work,aamaximal
maximalstrain
strainofof3128
3128µε induceda adeformation
µεinduced deformation ofof 0.822
0.822 mm.
mm.

Time Bulge Time

Controller 1 Chamber Controller 2
Samples

N2 Out
N2 In

Sealed Box

N2 In N2 Out
Resistance
Measurement

(a)
Figure 2. Cont.
Coatings 2020, 10, x FOR PEER REVIEW 5 of 16
Coatings 2020, 10, 792 5 of 15
Coatings 2020, 10, x FOR PEER REVIEW 5 of 16

(b)
(b)
Figure 2. (a) Bulge test fixture; (b) schematic of geometry relationship for the bulge test.
Figure2.2.(a)
Figure (a)Bulge
Bulgetest
testfixture;
fixture;(b)
(b)schematic
schematicof
ofgeometry
geometryrelationship
relationshipfor
forthe
thebulge
bulgetest.
test.
3. Results and Discussion
3.3.Results
Resultsand
andDiscussion
Discussion
3.1. Characteristics
3.1. Characteristics of of Inkjet-Printed
Inkjet-Printed CNTs
CNTs
3.1. Characteristics of Inkjet-Printed CNTs
Figure 33 shows
Figure shows thethe micro-Raman
micro-Raman spectroscopy
spectroscopy results
results ofof the
the CNT
CNT film.
film. Three
Three peaks
peaks of of typical
typical
CNT Figure
film 3 shows were
signatures the micro-Raman
observed. The spectroscopy
D band at results
1343 cm
−1 −1of the CNT the
represents film.defect
Threeband,
peaks theofGtypical
band
CNT film signatures were observed. The D band at 1343 cm −1 represents the defect band, the G band at
CNT
at 1587 film
cm signatures
−1 represents were
the observed.
graphitic The
band, D band
and theat 1343
peak atcm2692represents
cm the
−1 representsdefect
the band,
G’ band. the
TheG band
I D/IG
−1
1587 cm represents the graphitic band, and the peak at 2692 cm −1 −1 represents the G’ band. The ID /IG
at 1587
ratio was cmfound
−1 represents
to be the graphitic
be 0.735,
0.735, isband,
which is and
larger thanthethat
peakofat 2692
the cm represents
pristine CNTsbecause the G’the
because band. The ID/IG
ball-milling
ratio was found to which larger than that of the pristine CNTs the ball-milling
ratio was
processused found
usedto to
toobtainbe 0.735,
obtaineffective which is
effectivedispersionlarger than
dispersionincreases that
increasesthe of the
thenumber pristine
numberofofdefects CNTs because
defects[29].
[29]. the ball-milling
process
process used to obtain effective dispersion increases the number of defects [29].

G-Band
1587
G-Band
Units)

1587
ID/IG = 0.735
Units)

D-Band
1343
D-Band ID/IG = 0.735
1343
Intensity
Intensity (Arb.
(Arb.

1000 1500 2000 2500 3000

1000 1500 2000 (cm-12500
Raman Shift ) 3000
Raman Shift (cm-1)
Figure 3. Raman measurements for inkjet-printed CNT film.
Figure 3. Raman measurements for inkjet-printed CNT film.
Figure 3. Raman measurements for inkjet-printed CNT film.
Multi-pass printing is required for optimal thickness to improve conductivity and uniformity.
Multi-pass
The common printing
printing is required
direction for optimalprinting)
(unidirectional thicknessistoalong
improve conductivity
the pattern andtouniformity.
from left right and
Multi-pass printing is required for optimalprinting)
thicknessistoalong
improve conductivity and uniformity.
from top to bottom. Figure 4a shows an eight-layer CNT film with unidirectional printingto
The common printing direction (unidirectional the pattern from left right
from leftand
to
The
from common
top to printing
bottom. direction
Figure 4a (unidirectional
shows an printing)
eight-layer CNT is
filmalong
with the pattern
unidirectionalfrom left to
printing right
from and
left
right. The CNTs were agglomerated on the left edge because of the edge-enhanced evaporation of
from
to topThe
right. to bottom.
CNTs wereFigure 4a shows an
agglomerated oneight-layer
the left edgeCNT film with
because of theunidirectional printing from left
the aqueous multi-walled CNT ink, a phenomenon referred to as the edge-enhanced
coffee ring effect.evaporation
More voids of
to right.
the aqueous The CNTs were agglomerated on the left edge because of the edge-enhanced evaporation of
were found atmulti-walled
the right edge.CNTTheink, a phenomenon
density of CNTs was referred to asfrom
decreased the coffee
the leftring effect.
to right More
side. voids
In order
the aqueous multi-walled CNT ink, a phenomenon referred to as the coffee ring
were found at the right edge. The density of CNTs was decreased from the left to right side. In order effect. More voids
were found at the right edge. The density of CNTs was decreased from the left to right side. In order
Coatings 2020, 10, 792 6 of 15

Coatings 2020, 10, x FOR PEER REVIEW 6 of 16

to improve coffee ring effect, multi-pass printing with pattern rotation of inkjet-printed CNT films
to adopted.
was improve coffee ring effect, multi-pass printing with pattern rotation of inkjet-printed CNT films
wasFigure
adopted.4b,c show CNT films rotated by 180◦ and 90◦ after printing for every 4 and 2 passes,
Figure 4b,c show CNT
respectively. Comparing films morphologies,
the surface rotated by 180°the andCNT90°film
after printing
printed for180
with every 4 andafter
◦ rotation 2 passes,
every
respectively.
2 passes featuredComparing the surfaceon
slight agglomeration morphologies,
the left and rightthe sides.
CNT film
The printed
densitieswith 180°on
of CNTs rotation
the topafter
and
every 2 passes featured slight agglomeration on the left and right sides.
◦ The densities
bottom sides was lower than those on the left and right sides. For 90 rotation, the densities of CNTs of CNTs on the
top and bottom sides was lower than those on the left and right sides. For 90°
for all sides were similar. It indicates good uniform was obtained for the CNT film printed with 90◦ rotation, the densities
of CNTsFigure
rotation. for all5sides were similar.
compares the sheetIt resistivities
indicates good anduniform was of
thicknesses obtained for the
eight-layer CNTCNT filmobtained
films printed
by unidirectional printing, 180 rotation, and 90 rotation. The thickness was measured from films
with 90° rotation. Figure 5 compares
◦ the sheet resistivities
◦ and thicknesses of eight-layer CNT SEM
obtained byfor
cross-section unidirectional
three pointsprinting,
from left180° rotation,
to right side.and
The 90° rotation.
error bar ofThe thickness
thickness was measured
decreases becausefrom
the
SEM cross-section for three points from
surface roughness decreases by pattern rotation. left to right side. The error bar of thickness decreases because
theThe
surface
sheetroughness
resistance decreases by pattern
also decreased withrotation.
a decrease in the rotation degree. This indicates that
The sheet resistance also decreased with a decrease in the rotation degree. This indicates that
pattern rotation improved the CNT film uniformity. The thickness of eight-layer CNT film was about
pattern rotation improved the CNT film uniformity. The thickness of eight-layer CNT film was about
1.5 µm for 90◦ rotation sample. The thickness and density of CNT film for unidirectional printing, 180◦
1.5 µm for 90° rotation sample. The thickness and density of CNT film for unidirectional printing,
rotation, and 90◦ rotation are consistent with sheet resistivities. The pattern rotation of CNTs provides
180° rotation, and 90° rotation are consistent with sheet resistivities. The pattern rotation of CNTs
uniform films, leading to lower and stable resistance.
provides uniform films, leading to lower and stable resistance.

(a)

(b)

Figure 4. Cont.
 Coatings 2020, 10, 792 7 of 15
Coatings 2020, 10, x FOR PEER REVIEW 7 of 16
Coatings 2020, 10, x FOR PEER REVIEW 7 of 16

(c)(c)
Figure 4. 4.
Figure Optical
Opticalphotographs
photographs and
andsurface
surface morphologies
morphologies ofofeight-layer
eight-layer CNT film
CNT(a)film with
with (a)
(a)
Figure 4. Optical photographs and surface morphologies of eight-layer CNT film with unidirectional
unidirectional printing;
unidirectional ◦ (b) 180° rotation;
◦ (c) 90° rotation.
printing; (b) 180 printing;
rotation;(b)
(c)180°
90 rotation;
rotation. (c) 90° rotation.

3.53.5 2.0
2.0
Sheet resistance (kΩ/sq.)
Sheet resistance (kΩ/sq.)

3.03.0
1.5
1.5
2.52.5
Thickness (μm)
2.02.0 1.0
1.0 Thickness (μm)

1.51.5
0.5
0.5
1.01.0 8-layer
8-layer
o
Baked @@
Baked 150
150C,oC,
3 h3 h
0.50.5 0.0
0.0
unidirectional
unidirectional180-rotation
180-rotation 90-rotation
90-rotation
Figure
Figure
Figure
Sheetresistivities
5.5.Sheet resistivities
5. Sheet resistivities
and
and thicknesses
thicknesses
and ofof
thicknesses
eight-layer
eight-layer
of
CNT
CNT
eight-layer
films
CNTfilms
for
filmsfor
by unidirectional
forbybyunidirectional
printing,
unidirectionalprinting,
printing,
180 ◦ rotation, and 90◦ rotation.
180°
180° rotation,
rotation, andand
90°90° rotation.
rotation.

Figure 6 illustrates sheet resistance of 90◦ -rotated CNT films with respect to the number of printing
Figure
Figure 6 illustrates
6 illustrates sheet
sheet resistanceofof90°-rotated
resistance 90°-rotatedCNT CNTfilms
filmswith withrespect
respecttotothe thenumber
numberof of
passes. The number of every printing pass was calculated by dividing the total number of passes by
printing passes. The number of every printing pass was calculated
printing passes. The number of every printing pass was calculated by dividing the total number of by dividing the total number of
4. As expected, the sheet resistance decreased as the number of passes increased. A saturated sheet
passes
passes by by
4. 4.AsAs expected,
expected, thethe sheetresistance
sheet resistancedecreased
decreasedasasthe thenumber
numberofofpassespassesincreased.
increased. AA
resistance
saturated
was
sheet
found after 20
resistance was
layers.
found
The
after
sheetlayers.
resistance of theresistance
CNT filmofwas formed bywasthousands
saturated sheet resistance was found after 2020layers. The The sheetresistance
sheet of thetheCNT
CNTfilm filmwas formed
formed
ofbyindividual
thousands CNTs
of through
individual van
CNTs der Waals
through forces.
van der The
Waalsincrease
forces.in the
The number
increase of
in CNT
the passes
number ofmeans
CNT
by thousands of individual CNTs through van der Waals forces. The increase in the number of CNT
anpasses
increase in
means the CNTs density; therefore, the sheet resistance decreased with the number of passes.
passes means anan increase
increase in in
thethe CNTs
CNTs density;therefore,
density; therefore,the thesheet
sheetresistance
resistancedecreased
decreasedwith withthe the
The saturated
number sheet resistance
of passes. The saturatedaftersheet
20 layers indicates
resistance after the more connection points
morebetween individual
number of passes. The saturated sheet resistance after 2020 layersindicates
layers indicatesthe themore connection
connection points
points
CNTs.
between Additionally,
individual sheet
CNTs. resistance
Additionally, uniformity
sheet of the CNT
resistance films relies
uniformity on CNT
the number of printing
between individual CNTs. Additionally, sheet resistance uniformity ofofthe
theCNT films filmsrelies
relies ononthe
the
passes,
number as shown
of printingin the error
passes, bar. To analyze the surface uniformity, the surface morphologies of CNT
number of printing passes, as as shown
shown in in
thethe error
error bar.
bar. ToToanalyze
analyzethe thesurface
surfaceuniformity,
uniformity,the thesurface
surface
films at magnifications of 7000× from 8- to 20-layer printing are displayed in Figureare 7. A significant
morphologies of CNT films at magnifications of 7000× from 8- to 20-layer printing are displayed inin
morphologies of CNT films at magnifications of 7000× from 8- to 20-layer printing displayed
number
Figure of
7. voids occurrednumber
A significant on the 8-pass-printed film;onthethevoids became less film; pronounced onbecame
the 20-layer
Figure 7. A significant number of of voids
voids occurred
occurred on the 8-pass-printed
8-pass-printed film; thethevoids
voids became less less
film, resulting on in lower sheet resistance. Thein 20-layer film featured a uniform surface, resulting in
pronounced on the 20-layer film, resulting in lower sheet resistance. The 20-layer film featured aa
pronounced the 20-layer film, resulting lower sheet resistance. The 20-layer film featured
saturated
uniform resistance.
surface, The
resulting decreased sheet resistance is attributed to the roughness-to-thickness ratio
uniform surface, resulting in in saturated
saturated resistance.
resistance. The
The decreasedsheet
decreased sheetresistance
resistanceisisattributed
attributedtotothe the
decrease with an increase
roughness-to-thickness in the
ratio number
decrease withofan passes.
increase Thein 20-layer
the number printing
of wasThe
passes. selected
20-layer to printing
provide
roughness-to-thickness ratio decrease with an increase in the number of passes. The 20-layer printing
appropriate
was selecteduniformity
to provide and conductivity.
appropriate uniformity and conductivity.
was selected to provide appropriate uniformity and conductivity.
Coatings 2020, 10, 792 8 of 15
Coatings 2020,
Coatings 10,10,
2020, x FOR PEER
x FOR REVIEW
PEER REVIEW 8 of8 16
of 16

1.51.5 o o
Baked @@
Baked 150150
C, C,
3 h3 h

Sheet Resistance (kΩ/sq.)

Sheet Resistance (kΩ/sq.)
1.21.2

0.90.9

0.60.6

0.30.3

0.00.0
88 1212 1616 2020 2424
No. of passes printed
No. of passes printed
Figure Sheet
6. Sheet resistance ofof ◦ -rotated CNT films with respect to the number of printing passes.
9090°-rotated
Figure 6. resistance
Sheet of
resistance 90°-rotated CNT films
CNT with
films respect
with respect to the number of printing passes.

8-layer
8-layer 12-layer
12-layer

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

16-layer
16-layer 20-layer
20-layer

(c)(c) (d)(d)
Figure 7. Surface
Figure 7.Surfacemorphologies
Surface morphologies
morphologies ofofof
CNT films
CNT
CNT at at
films
films magnifications
at of of
magnifications
magnifications 7000× from
7000×
7000× from(a)(a)
from 8-layer printing;
(a)8-layer
8-layer (b)(b)
printing;
printing;
12-layer printing;
(b)12-layer
12-layer (c)(c)
printing;
printing; 16-layer printing;
(c)16-layer
16-layer (d)(d)
printing;
printing; 20-layer
(d) printing.
20-layer
20-layer printing.
printing.

3.2.
3.2. Electrical
Electrical
3.2. Properties
Properties
Electrical of of
of
Properties Strain
Strain Sensors
Sensors
Strain Sensors
ToTo
To investigate
investigate
investigate the sensor
thethe
sensor
sensor repeatability,
repeatability,
repeatability, multiple
multiple
multiple cycles
cycles
cycleswere
were
were carried
carried
carriedout
outoutvia the
viavia bulge
thethe
bulge
bulge test
test using
using
test using
cyclic
cyclic stretching
stretching
cyclic stretchingand
and relaxation
andrelaxation
relaxation for
forthe
the20-layer
for the20-layerfully
20-layer inkjet-printed
fully
fullyinkjet-printed strain
inkjet-printed sensor.
strain
strainsensor.TheThe
sensor. sensing
The areaarea
sensing
sensing had
area
a length
had
had a of
a length 3 mm
length 3and
of of mm a and
3 mm width
and of 1 mm.
a width
a width Figure
of of
1 mm.
1 mm. 8Figure
shows
Figure 8the transient
shows
8 shows thethe response
transient
transient ofresponse
the relative
response of of resistance
the
therelative
relative
change
resistance(∆R/R
resistance 0 ) for(ΔR/R
change
change the 20-layer
(ΔR/R 0) 0for
) for strain
thethe sensor
20-layer
20-layer under
strain
strain repeated
sensor
sensor understretching/releasing
under repeated
repeated cycles with tensile
stretching/releasing
stretching/releasing cycles
cycles
with
withtensile strain
tensile strainfrom
from 0 to
0 to3128
3128µε,µε,
whereby
whereby each
eachstrain
strainwas
washeld
heldandandreleased
released forfor
10,10,
30,30,
andand6060s. s.
Coatings 2020, 10, 792 9 of 15

strain from 0 to 3128 µε, whereby each strain was held and released for 10, 30, and 60 s. The parameter
R0 is the initial resistance before strain. The resistance increased when tensile strain was applied
because the CNTs were disconnected. The resistance could not return to its original value when the
first strain was released because the CNTs network was permanently destroyed at the first strain,
which resulted in resistance increase. This irrecoverable resistance is attributed to some irreversible
deformation. The ∆R/R0 was almost recovered to first release for all cycles, excluding the permanent
destruction, indicating that the strain sensor possessed better recoverability characteristics. In order
to provide a stable baseline for strain, we first measured the 10 maximum strain cycles to be tested
to exclude permanent damage. This setting was used in subsequent experiments. Then, the relative
change in ∆R/R0 between the strain and subsequent release values (|∆R/R0 |) is the quantification of
the number of ∆R/R0 . This value is calculated to exclude continue damage of the CNTs network
and substrate recovery rate after strain, as a comparison of strain strength. The average |∆R/R0 |
for 10 cycles were 0.99%, and 1.05% after a strain holding/release time of 30 and 60 s, respectively.
The |∆R/R0 | were almost similar after a strain holding/release time of 30 and 60 s, and the resistance was
maintained at a constant value over a long time period (60 s), demonstrating the stability performance.
However,
Coatings 2020, |∆R/R
the10, 0 | PEER
x FOR was REVIEW
0.93% after a strain holding/release time of 10 s. Although the strain 10
rate
of 16
was considerably high for 10 s, the release rate was slightly low, resulting in a lower |∆R/R0 |. Therefore,
we chose 30 s in subsequent experiments.

2.5

2.0

1.5
ΔR/R0 (%)

1.0

0.5
|ΔR/R0|
R0
0.0 W=1 mm, L=3 mm
20-layer & 90° rotation @ Strain=3128 με
-0.5
0 40 80 120 160 200 240
Time (s)
(a)

2.5

2.0

1.5
ΔR/R0 (%)

1.0

0.5
R0
0.0 W=1 mm, L=3 mm
20-layer & 90° rotation @ Strain=3128 με
-0.5
0 1 2 3 4 5 6 7 8 9 10 11 12
Time (min)
(b)
Figure 8. Cont.
2.5

2.0

1.5
(%)
 0.0 W=1 mm, L=3 mm
20-layer & 90° rotation @ Strain=3128 με
-0.5
0 1 2 3 4 5 6 7 8 9 10 11 12
Coatings 2020, 10, x FOR PEER REVIEW Time (min) 10 of 16
Coatings 2020, 10, 792 10 of 15
(b)

2.5
2.5
2.0
2.0
1.5
1.5

0 0 (%)
(%)
1.0

ΔR/R
1.0
ΔR/R 0.5
0.5
R 0
0.0 R W=1 mm, L=3 mm
0
0.0 20-layer
W=1 mm, & L=3 rotation @ Strain=3128 με
90° mm
-0.5 20-layer & 90° rotation @ Strain=3128 με
-0.5 0 1 2 3 4 5 6 7 8 9 10 11 12
0 2 4 6 8Time 10 (min)
12 14 16 18 20 22
Time (min)
(b)
(c)
Figure 8. Transient
Figure 8. Transient 2.5 of the
response
response relative
of the resistance
relative resistancechange (∆R/R
change 0 ) for
(ΔR/R thethe
0) for 20-layer strain
20-layer sensor
strain sensor
under repeated
under stretching/releasing
repeated stretching/releasing cycles
cycleswith tensile
with strain
tensile from
strain 0 to0 3128
from whereby
µε,µε,
to 3128 each
whereby strain
each strain
waswas
held andand
held released for2.0
released (a) 10
for (a) s;10(b) 30 s;30(c)
s; (b) s; 60
(c) s.60 s.

To examine the strain1.5 performance of the sensor excluding the permanent damage, 10 strain
ΔR/R0 (%)

cycles with tensile strain from 0 to 3128 µε were performed before the strain sensor performance was
evaluated. Figure 9a shows1.0 the stretch/release transient response as a function of various strains from
71 to 3128 µε for 20-layer strain sensor. ∆R/R0 was increased with tensile strain increased. ∆R/R0 after
0.5
strain release was slightly increased after 3128 µε. After the maximum strain was released, ∆R/R0 after
R
the subsequent strain was 0.0 decreased with strain decreased and could be restored to R0 after 456 µε
0

W=1 mm, L=3 mm

because the substrate recovery was slower than CNT. Figure 9b summarizes the stretch/release curves
20-layer & 90° rotation @ Strain=3128 με
of |∆R/R0 | and ∆R/R0 for 20-layer
-0.5 strain sensor. As shown, |∆R/R0 | and ∆R/R0 increased almost linearly
with an increase in the tensile 0strain.
2 The
4 6 8 10
release curve12 of14 16 18
∆R/R 20 22
0 is slightly larger than stretch curve
Time (min)
due to the slow recovery of substrate. The |∆R/R0 | values of the stretch/release curves were almost
similar, indicating that the CNT film can be fully reversible
(c) by excluding substrate recovery. The GF is
the ratio of fractional change in resistance versus the mechanical strain and is given by [30]:
Figure 8. Transient response of the relative resistance change (ΔR/R0) for the 20-layer strain sensor
∆R strain from 0 to 3128 µε, whereby each strain
under repeated stretching/releasing cycles with tensile
GF = /ε (2)
was held and released for (a) 10 s; (b) 30 s; (c) 60 s.R

1.5
W=1 mm, L=3 mm, 20-layer & 90° rotation
Strain=3128 με

2549
1.0 2549
ΔR/R0 (%)

1944
1944
1370
1370
0.5 586
586
456 456
71 71
R0
0.0 |ΔR/R0|

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time (min)

(a)

Figure 9. Cont.
Coatings
Coatings2020,
2020,10,
10,x 792
FOR PEER REVIEW 1111
ofof
1615

1.2 2.0
W=1 mm, L=3 mm, 20-layer & 90° rotation
1.8
1.0 1.6
0.8 1.4

|ΔR/R0| (%)

ΔR/R0 (%)
1.2
0.6 release stretch 1.0
0.8
release
0.4 0.6
stretch

0.2 0.4
0.2
0.0 0.0
0 500 1000 1500 2000 2500 3000 3500
Strain (με)

(b)
Figure9.9.(a)(a)Stretch/release
Figure Stretch/releasetransient
transientresponse
responseasasa afunction
functionofofvarious
variousstrains
strains from7171toto3128
from 3128µεµεfor
for
20-layerstrain
20-layer strainsensor;
sensor;(b)
(b)stretch/release
stretch/releasecurves
curvesofofthe
the20-layer
20-layerstrain
strainsensor.
sensor.

From 10
Figure theshows
slopethe of the fitted line in
stretch/release Figure 8b,
transient |∆R/R0 | and
the GFasofa function
response ∆R/R0 strains
of various were calculated
from 71 toto
about
3128 µε 2.77 and 3.42, respectively.
for eight-layer strain sensor with The unidirectional
strain sensitivities printingof inkjet-printed
and 180° rotation. 20-layer strain sensor is
The unidirectional
comparable to the GF of commercial strain gauges [31,32]. The
printing eight-layer strain sensor causes damage after tensile of 71 µε. The relative resistance strain from 71 to 3128 µε could wasbe
clearly
larger distinguished
than 3% for tensile by the strain
strain from sensor,
456 to due3128to highly linear behavior.
µε. However, it is not the permanent destruction
Figure 10 shows the stretch/release transient response
because the relative resistance returned to R0 after each strain and the reversibility as a function of variousofstrains
71 µε from
was also71 to
3128 µε for eight-layer strain sensor with unidirectional printing and 180 ◦ rotation. The unidirectional
found. The relative resistance increases significantly for eight-layer strain sensor with unidirectional
printingwhen
printing eight-layer
tensilestrain
strainsensor
lager causes
than 456 damage
µε. The after tensileinofresistance
increase 71 µε. Thefor relative resistance
eight-layer was
strain larger
sensor
than 3% for tensile strain from 456 to 3128 µε. However, it is not
is due to the CNT density on the one side by unidirectional printing is lower, resulting in more the permanent destruction because
the relative resistance
disconnection of the CNTs returned
network to Rand0 after
pooreach strain andwith
connectivity the reversibility
silver film. For of 71 was alsoeight-
µε rotation
180° found.
The strain
layer relative resistance
sensors, the increases significantly
relative resistance couldfornoteight-layer
recover to strain
R0 and sensor with unidirectional
the negative printing
relative resistance
when
was tensile
found when strain lager
tensile thanlower
strain 456 µε. thanThe increase
456 µε. This inisresistance
because strainfor eight-layer
is appliedstrain
to thesensor
uneven is due
CNTto
the CNT
film density
resulting in on the one sidestrain
non-uniform by unidirectional
distributionprinting is lower, resulting
and permanent in more
destruction. disconnection
Some resistanceof
the CNTs
Coatings network
2020, 10, x and
FOR poor
PEER connectivity
REVIEW with silver film. For 180 ◦ rotation eight-layer strain sensors, 12 of 16
singularities were measured at the moment of stretching/releasing strain because of poor connection
the relative resistance could not recover to R and the negative
for uneven CNT film. Non-uniform CNT film 0causes damage when strain is applied. Therefore, the relative resistance was found when
lowerstrain
tensile
uniformity than
of thethat
lowerCNTofthan
20-layer
film456 because
µε.
cannot This of because
is
be ignored lower fordensity.
strain
the isFigure
applied
electronic 11bto
and shows the stretch/release
the uneven
mechanical CNT film resulting
properties. transient
in response
non-uniform
Figure 11a as shows
a strain
function
the of various
distribution
transient strains
and
response from
permanent
of 71destruction.
to 3128
the relative µε for
resistanceSomeeight-layer
resistance
change strain
for the sensor with
singularities
eight-layer were90°
with
90° rotation.
measured ΔR/R
at thesensor
rotation strain was
moment
0 also increased with
of stretching/releasing
under tensile
repeated stretching/releasing strain increased
strain because and
of poor
cycles with was decreased
connection
tensile strain from after release.
for 0uneven
to 3128 CNTµε,The
strain
whereby behaviors
film. Non-uniform
each strain was including
CNTheld film andreversibility,
causes damage
released reproducibility,
forwhen
30 s. Thestrain and stability
is applied.
resistance for
alsoTherefore, 8-layer
increasedthe when strain
uniformity sensor was
of the
tensile strain
CNT
was similar
film to
applied 20-layer
cannot
and could strain
be ignored sensor.
for thetoelectronic
not recover its original and mechanical
value, which was properties.
similar to 20-layer strain sensor.
The average |ΔR/R0| of 8-layer strain sensor for 10 cycles were 0.76% for a strain holding/release time
5.5
of 30 s. The irrecoverable relative resistance
W=1 mm, L=3change
mm, 8-layerand average |ΔR/R
& unidirectional 0| of 8-layer strain sensor is
printing
5.0
lower than that of 20-layer because of lower density. Figure
Strain=3128 με 11b shows the stretch/release transient
4.5
2549
response as a function of various
4.0 strains from 1944712549
to 3128 µε for eight-layer strain sensor with 90°
1944
1370 1370
rotation. ΔR/R0 was also increased
3.5 with tensile
586
strain increased and 586was decreased after release. The
ΔR/R0 (%)

3.0
strain behaviors including reversibility, reproducibility, and stability 456 for 8-layer strain sensor was

2.5
similar to 20-layer strain sensor.
2.0
1.5 456

1.0 71
71
0.5
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time (min)
(a)
Figure 10. Cont.
1.5
W=1 mm, L=3 mm, 8-layer & 180° rotation

Strain=3128 με
2549
1.0 2549 1944
(%)

1370
1944 586
ings 2020, 10, x FOR PEER REVIEW 1.0 71 12 of 16
71
0.5
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
5.5
W=1 mm, L=3 mm, 8-layer & unidirectional printing Time (min)
5.0
Coatings 2020, 10, 792 Strain=3128 με 12 of 15
4.5 (a)
2549 2549
4.0 1944
1944
1370 1370
3.5 586 586
ΔR/R0 (%) 1.5
3.0 W=1 mm, L=3 456 mm, 8-layer & 180° rotation
2.5
Strain=3128 με
2.0 2549
1.5 456 1.0 2549 1944

ΔR/R0 (%)
1.0 71
1370
1944
71 586
0.5 456
0.0 0.5
1370
71
0 1 2 3 4 5 6 7 8 9 10 11 586 12 13 14

Time (min)71 456

0.0
(a)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1.5
W=1 mm, L=3 mm, 8-layer & 180° rotation
Time (min)
Strain=3128 με
(b)
2549
1.0 1944
2549transient
Figure Stretch/release
10.10.
Figure Stretch/release transientresponse
responseasas
a function
a functionof of
various
variousstrains from
strains 7171
from to to
3128 forfor
µεµε
3128
ΔR/R0 (%)

1370
eight-layer strain sensor with (a)(a)
unidirectional printing; (b)(b) ◦ rotation.
180180°
eight-layer strain sensor
1944 with unidirectional
586 printing; rotation.
456
1370
0.5 11a shows the transient response of71the relative resistance change for the eight-layer with
Figure
586
90◦ rotation strain
456
sensor under
2.5 repeated stretching/releasing cycles with tensile strain from 0 to 3128
W=1 mm, L=3 mm
µε, whereby each strain was held8-layer
71 and released for 30 s. The resistance also increased when tensile strain
& 90° rotation @ Strain=3128 με
0.0
was applied and could not 2.0recover to its original value, which was similar to 20-layer strain sensor.
The average |∆R/R0 | of 8-layer strain sensor for 10 cycles were 0.76% for a strain holding/release time of
0 1 2 3 4 5 6 1.5 7 8 9 10 11 12 13 14
ΔR/R0 (%)

30 s. The irrecoverable relative resistance change and average |∆R/R0 | of 8-layer strain sensor is lower
Time (min)
than that of 20-layer because1.0 of lower density. Figure 11b shows the stretch/release transient response
(b)
as a function of various strains from 71 to 3128 µε for eight-layer strain sensor with 90◦ rotation. ∆R/R0
0.5
was alsotransient
Figure 10. Stretch/release increased with tensile
response strain increased
as a function of various and was
strains decreased
from after
71 to 3128 µε release.
for The strain behaviors
R 0
including reversibility, 0.0
reproducibility,
Coatings and
2020, 10, stability
x FOR
eight-layer strain sensor with (a) unidirectional printing; (b) 180° rotation. PEER for 8-layer
REVIEW strain sensor was similar to 20-layer
strain sensor.
-0.5
0 1 2 3 4 5 6 7 8 9 10 11 12
2.5 1.5
Time (min) W=1 mm, L=3 mm, 8-layer & 90° rotation
W=1 mm, L=3 mm
8-layer & 90° rotation @ Strain=3128 με (a)
2.0
1.0 Strain=3128 με
1.5
ΔR/R0 (%)
ΔR/R0 (%)

2549
2549
1944
1.0 1944
1370
0.5 1370
586
0.5 586
456
456
71 71
R0
0.0 R0
0.0 |ΔR/R0|

-0.5
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time (min) Time (min)
(a) (b)
Figure
Figure 11. Stretch/Release transient 11. Stretch/Release
response of the relativetransient response
resistance change of
forthe relative resistance
eight-layer strain change for eight-laye
sensor with 90◦ rotation (a) under
sensor with stretching/releasing
repeated 90° rotation (a) under repeated
cycles stretching/releasing
with tensile strain from 0 cycles
to with tensile strain fr
3128 µε, whereby each strain was3128
heldµε,
andwhereby
releasedeach strain
for 30 s; (b)was held and
at various released
strains from for7130tos;3128
(b) at various strains from 71 to
µε.

Figure 12 compares the stretch/release curves ofthe

Figure 12 compares |∆R/R 0 | for eight-layer
stretch/release curves strain sensor
of |ΔR/R by eight-layer strain
0| for
◦
unidirectional printing, 180 unidirectional ◦
rotation, andprinting,
90 rotation with tensile
180° rotation, and strain from 71
90° rotation withto tensile
3128 µε.
strain from 71 to 3
The unidirectional printing strain sensor can only apply a strain of 71 µε. The non-linear behavior of
unidirectional printing strain sensor can only apply a strain of 71 µε. The non-linear
|∆R/R0 | were found after 456 |ΔR/R
µε. The0| strain sensitivities
were found for µε.
after 456 eight-layer
The strainstrain sensor with
sensitivities 180◦ and strain sensor w
for eight-layer
90° rotation exhibited a linear and reversible behavior under 71 to 3128 µε. However,
was negative while tensile strain lower than 456 µε. It indicates the strain sensor with 1
can only measure the tensile strain in the range of 586 to 3128 µε. The GF for 180°- and
strain sensor were about 1.85 and 1.93, respectively. The lower GF and smaller measur
180° rotation strain sensor was due to uneven CNT film. The strain distribution in an un
Coatings 2020, 10, 792 13 of 15

90◦ rotation exhibited a linear and reversible behavior under 71 to 3128 µε. However, the |∆R/R0 |
was negative while tensile strain lower than 456 µε. It indicates the strain sensor with 180◦ rotation
can only measure the tensile strain in the range of 586 to 3128 µε. The GF for 180◦ - and 90◦ -rotated
strain sensor were about 1.85 and 1.93, respectively. The lower GF and smaller measured range of 180◦
rotation strain sensor was due to uneven CNT film. The strain distribution in an unidirectional printing
and 180◦ rotation strain sensors was non-uniform because of uneven CNT film, which decreases the
linearity, sensitivity and measurement range of the sensor. It indicates that the uniformity of CNT film
affects the mechanical properties. The uniformity of the CNT film was closely related to the electrical
and mechanical properties. In addition, the GF of eight-layer strain sensor was lower than that of
20-layer strain sensor because of lower CNT density and higher resistance. Higher multi-pass printing
provides a uniform surface resulting in better strain behaviors, which was consistent with surface
morphologies and sheet resistances. The results demonstrate the mechanical deformation and electrical
resistance of the fully inkjet-printed CNT-based strain sensor; the sensor features high reversibility,
reproducibility, and stability to afford microstrain detection and classification. Fully inkjet-printed CNT
film with multi-pass printing and pattern rotation provides better and stable mechanical performance
Coatings
for strain2020, 10, x FOR PEER REVIEW
applications. 14 of 16

4
W=1 mm, L=3 mm, 8-layer

3
Stretch Release
|ΔR/R0| (%)

Unidirectional Printing
180° rotation
2 90° rotation

0
0 500 1000 1500 2000 2500 3000 3500
Strain (με)

Figure
Figure12.12.Stretch/release
Stretch/releasecurves of the
curves of eight-layer strainstrain
the eight-layer sensors by unidirectional
sensors 180◦ rotation,
printing,printing,
by unidirectional 180°
and ◦
90 rotation.
rotation, and 90° rotation.

4. Conclusions
4. Conclusions
A fully inkjet-printed strain sensor was successfully fabricated using a CNT-sensing film and a
A fully inkjet-printed strain sensor was successfully fabricated using a CNT-sensing film and a
silver conductive film. Based on the digital image of the inkjet printing technology, a uniform and
silver conductive film. Based on the digital image of the inkjet printing technology, a uniform and
stable-resistance CNT film was obtained via multi-pass printing and pattern rotation. Inkjet-printed
stable-resistance CNT film was obtained via multi-pass printing and pattern rotation. Inkjet-printed
CNT and silver films were fabricated layer-by-layer to form strain sensors. The uneven carbon nanotube
CNT and silver films were fabricated layer-by-layer to form strain sensors. The uneven carbon
film causes poor electrical and mechanical properties of strain sensors. The low strain behaviors such
nanotube film causes poor electrical and mechanical properties of strain sensors. The low strain
as non-linear, small measurement range, and low sensitivity were observed for strain sensor with
behaviors such as non-linear, small measurement range, and low sensitivity were observed for strain
unidirectional printing and 180◦ rotation because of uneven CNT film. The relative resistance change
sensor with unidirectional printing and 180° rotation because of uneven CNT film. The relative
for the 8- and 20-layer strain sensor with 90◦ rotation was 0.76% and 1.05% for 10 cycles between
resistance change for the 8- and 20-layer strain sensor with 90° rotation was 0.76% and 1.05% for 10
0 and 3128 µε strain, respectively. The GF of 8- and 20-layer strain sensors with 90◦ rotation were
cycles between 0 and 3128 µε strain, respectively. The GF of 8- and 20-layer strain sensors with 90°
1.93 and 2.77, respectively. The higher GF of 20-layer strain sensor was due to higher density and
rotation were 1.93 and 2.77, respectively. The higher GF of 20-layer strain sensor was due to higher
uniform surface of multi-pass printing. The strain sensors with 90◦ rotation exhibited a linear and
density and uniform surface of multi-pass printing. The strain sensors with 90° rotation exhibited a
reversible behavior under 71 to 3128 µε. The fully inkjet-printed CNT-based strain sensor showed
linear and reversible behavior under 71 to 3128 µε. The fully inkjet-printed CNT-based strain sensor
high reversibility, repeatability, and linearity; therefore, it can be applied to classify microstrain for
showed high reversibility, repeatability, and linearity; therefore, it can be applied to classify
distinguishing scleroderma hardness and thin-film crack detection.
microstrain for distinguishing scleroderma hardness and thin-film crack detection.

Author Contributions: Conceptualization, H.-L.K.; methodology, H.-L.K. and C.-L.C.; validation, C.-L.C. and
Y.-C.T.; resources, H.-L.K. and L.-C.C.; writing—original draft preparation, H.-L.K. and C.-L.C.; writing—review
and editing, L.-C.C., C.-B.C., and W.-H.C.; funding acquisition, H.-L.K., C.-B.C., and W.-H.C. All authors have
read and agreed to the published version of the manuscript.

Funding: This work was supported by the Chang Gung Memorial Hospital, Taiwan (CMRPD2H0311 &
BMRP957).
Coatings 2020, 10, 792 14 of 15

Author Contributions: Conceptualization, H.-L.K.; methodology, H.-L.K. and C.-L.C.; validation, C.-L.C.
and Y.-C.T.; resources, H.-L.K. and L.-C.C.; writing—original draft preparation, H.-L.K. and C.-L.C.; writing—
review and editing, L.-C.C., C.-B.C., and W.-H.C.; funding acquisition, H.-L.K., C.-B.C., and W.-H.C. All authors
have read and agreed to the published version of the manuscript.
Funding: This work was supported by the Chang Gung Memorial Hospital, Taiwan (CMRPD2H0311 & BMRP957).
Acknowledgments: The authors thank the Centre for Reliability Sciences and Technologies at Chang Gung
University for their help.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Kanoun, O.; Müller, C.; Benchirouf, A.; Sanli, A.; Dinh, T.N.; Al-Hamry, A.; Bu, L.; Gerlach, C.; Bouhamed, A.
Flexible carbon nanotube films for high performance strain sensors. Sensors 2014, 14, 10042–10071. [CrossRef]
[PubMed]
2. Slepchenkov, M.M.; Glukhova, O.E. Mechanical and electroconductive properties of mono- and bilayer
graphene–carbon nanotube films. Coatings 2019, 9, 74. [CrossRef]
3. Andrews, J.B.; Cardenas, J.A.; Lim, C.J.; Noyce, S.G.; Mullett, J.; Franklin, A.D. Fully printed and flexible
carbon nanotube transistors for pressure sensing in automobile tires. IEEE Sens. J. 2018, 18, 7875–7880.
[CrossRef]
4. He, Z.; Byun, J.-H.; Zhou, G.; Park, B.-J.; Kim, T.-H.; Lee, S.-B.; Yi, J.-W.; Um, M.-K.; Chou, T.-W. Effect
of MWCNT content on the mechanical and strain-sensing performance of Thermoplastic Polyurethane
composite fibers. Carbon 2019, 146, 701–708. [CrossRef]
5. Amjadi, M.; Yoon, Y.J.; Park, I. Ultra-stretchable and skin-mountable strain sensors using carbon nanotubes-
ecoflex nanocomposites. Nanotechnology 2015, 26, 375501. [CrossRef] [PubMed]
6. Yee, M.J.; Mubarak, N.M.; Khalid, M.; Abdullah, E.C.; Jagadish, P. Synthesis of polyvinyl alcohol (PVA)
infiltrated MWCNTs buckypaper for strain sensing application. Sci. Rep. 2018, 8, 17295. [CrossRef]
7. Bu, L.; Steitz, J.; Kanoun, O. Influence of processing parameters on properties of strain sensors based on
carbon nanotube films. In Proceedings of the International Multi-Conference on Systems, Signals and
Devices, Amman, Jordan, 27–30 June 2010.
8. Lee, D.; Park, C.W.; Min, N.K. Microfabrication of spray-coated SWCNT film strain gauges. In Proceedings
of the International Conference on Nanotechnology, Portland, OR, USA, 15–19 August 2011; pp. 166–171.
9. Sahatiya, P.; Badhulika, S. Eraser-based eco-friendly fabrication of a skin-like large-area matrix of flexible
carbon nanotube strain and pressure sensors. Nanotechnology 2017, 28, 095501. [CrossRef]
10. Chun, S.; Son, W.; Choi, C. Flexible pressure sensors using highly-oriented and free-standing carbon nanotube
sheets. Carbon 2018, 139, 586–592. [CrossRef]
11. Chen, S.; Li, Y.; Lin, Y.; He, P.; Long, T.; Deng, C.; Chen, Z.; Chen, G.; Tao, H.; Lan, L.; et al. Inkjet-printed
top-gate thin-film transistors based on InGaSnO semiconductor layer with improved etching resistance.
Coatings 2020, 10, 425. [CrossRef]
12. Cho, C.-L.; Kao, H.-L.; Chang, L.-C.; Wu, Y.-H.; Chiu, C.-H. Inkjet-printed vertical interconnects for ultrathin
system-on-package technology. Surf. Coat. Technol. 2019, 359, 85–89. [CrossRef]
13. Qin, Y.; Kwon, H.-J.; Subrahmanyam, A.; Howlader, M.M.R.; Selvaganapathy, P.R.; Adronov, A.; Deen, M.J.
Inkjet-printed bifunctional carbon nanotubes for pH sensing. Mater. Lett. 2016, 176, 68–70. [CrossRef]
14. Yan, X.; Shi, K.; Chu, X.; Yang, F.; Chi, Y.; Yang, X. Stepped annealed inkjet-printed InGaZnO thin-film
transistors. Coatings 2019, 9, 619. [CrossRef]
15. Schneck, M.; Gollnau, M.; Lutter-Günther, M.; Haller, B.; Schlick, G.; Lakomiec, M.; Reinhart, G. Evaluating
the use of additive manufacturing in industry applications. Procedia CIRP 2019, 81, 19–23. [CrossRef]
16. Jiménez, M.; Romero, L.; Dominguez, I.A.; Espinosa, M.D.M.; Dominguez, M. Additive manufacturing
technologies: An overview about 3D printing methods and future prospects. Complexity 2019, 2019, 1–30.
[CrossRef]
17. Assadi, H.; Gärtner, F. Compression test: A key step towards tailoring of feedstock powder for cold spraying.
Coatings 2020, 10, 458. [CrossRef]
Coatings 2020, 10, 792 15 of 15

18. Tang, Z.; Yao, D.; Hu, S.; Du, D.; Shao, W.; Tang, B.; Fan, J.-M.; Tang, X.-G.; Gao, J. Highly conductive, washable
and super-hydrophobic wearable carbonnanotubes e-textile for vacuum pressure sensors. Sens. Actuators
A Phys. 2020, 303, 111710. [CrossRef]
19. Srivastava, R.K.; Vemuru, V.S.M.; Zeng, Y.; Vajtai, R.; Nagarajaiah, S.; Ajayan, P.M.; Srivastava, A. The strain
sensing and thermal–mechanical behavior of flexible multi-walled carbon nanotube/polystyrene composite
films. Carbon 2011, 49, 3928–3936. [CrossRef]
20. Tortorich, R.P.; Choi, J.-W. Inkjet printing of carbon nanotubes. Nanomaterials 2013, 3, 453–468. [CrossRef]
21. Gutierrez, S.M.V.; He, Y.; Cao, Y.; Stone, D.; Walsh, Z.; Malhotra, R.; Chen, H.-L.; Chang, C.-H.; Robinson, S.C.
Feasibility and surface evaluation of the pigment from scytalidium cuboideum for inkjet printing on textiles.
Coatings 2019, 9, 266. [CrossRef]
22. Soum, V.; Park, S.; Brilian, A.I.; Kim, Y.; Ryu, M.Y.; Brazell, T.; Burpo, F.J.; Parker, K.K.; Kwon, O.-S.; Shin, K.
Inkjet-printed carbon nanotubes for fabricating a spoof fingerprint on paper. ACS Omega 2019, 4, 8626–8631.
[CrossRef]
23. Kim, T.; Song, H.; Ha, J.; Kim, S.; Kim, D.; Chung, S.; Lee, J.; Hong, Y. Inkjet-printed stretchable single-walled
carbon nanotube electrodes with excellent mechanical properties. Appl. Phys. Lett. 2014, 104, 113103.
[CrossRef]
24. Zhao, Y.; Beisteiner, C.; Gschossmann, S.; Schagerl, M. An inkjet-printed carbon nanotube strain distribution
sensor for quasi real-time strain monitoring of lightweight design materials. Adv. Sci. Technol. 2016, 101, 3–8.
[CrossRef]
25. Dinh, N.T.; Sowade, E.; Blaudeck, T.; Hermann, S.; Rodriguez, R.D.; Zahn, D.R.T.; Schulz, S.E.;
Baumann, R.R.; Kanoun, O. High-resolution inkjet printing of conductive carbon nanotube twin lines
utilizing evaporation-driven self-assembly. Carbon 2016, 96, 382–393. [CrossRef]
26. Michelis, F.; Bodelot, L.; Bonnassieux, Y.; Lebental, B. Highly reproducible, hysteresis-free, flexible strain
sensors by inkjet printing of carbon nanotubes. Carbon 2015, 95, 1020–1026. [CrossRef]
27. CLTE-MW™ Laminates. Available online: https://rogerscorp.com/-/media/project/rogerscorp/documents/
advanced-connectivity-solutions/english/data-sheets/clte-mw-laminates-data-sheet.pdf (accessed on 14
August 2020).
28. Muller, R.M.; Conragan, J. A Metal-insulator-piezoelectric semiconductor electromechanical transducer.
IEEE Trans. Electron. Devices 1965, 12, 590–595. [CrossRef]
29. Kwon, O.-S.; Kim, H.; Ko, H.; Lee, J.; Lee, B.; Jung, C.-H.; Choi, J.-H.; Shin, K. Fabrication and characterization
of inkjet-printed carbon nanotube electrode patterns on paper. Carbon 2013, 58, 116–127. [CrossRef]
30. Li, X.; Levy, C.; Elaadil, L. Multiwalled carbon nanotube film for strain sensing. Nanotechnology 2008,
19, 045501. [CrossRef]
31. Strain Gage Technical Data. Available online: https://www.omega.co.uk/techref/pdf/STRAIN_GAGE_
TECHNICAL_DATA.pdf (accessed on 14 August 2020).
32. Strain Gages. Available online: http://spectromas.ro/wp-content/uploads/2018/10/S01265.pdf (accessed on 14
August 2020).

© 2020 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/).