Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Polymer 82 (2016) 181e189

Contents lists available at ScienceDirect

Polymer
journal homepage: www.elsevier.com/locate/polymer

Strain sensing, electrical and mechanical properties of polycarbonate/


multiwall carbon nanotube monofilament fibers fabricated by melt
spinning
€tschke a, *, Harald Brünig a, Gert Heinrich a, b
 Roberto Bautista-Quijano a, b, Petra Po
Jose
a
Leibniz-Institut für Polymerforschung Dresden e.V. (IPF), Hohe Str. 6, 01069 Dresden, Germany
b
Technische Universita €t Dresden, Institut für Werkstoffwissenschaft, 01062 Dresden, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Polycarbonate (PC)/multiwall carbon nanotube (MWCNT)/composites were prepared by melt mixing and
Received 18 August 2015 subsequently melt spun to monofilament fibers. Electrical, mechanical and piezoresistive (strain sensing)
Received in revised form properties of the fibers were evaluated as a function of the MWCNT weight concentration, which was
9 November 2015
varied up to 6 wt%. After the incorporation of MWCNTs, a decrease in electrical resistivity of the fibers
Accepted 13 November 2015
Available online 1 December 2015
was found for MWCNT concentrations as low as 2 wt%. Furthermore, an important effect of the draw
down ratio (DDR) on the electrical properties of the fibers was observed. While the electrical percolation
threshold (4c) for the bulk material was <1 wt% and ~1 wt% for the undrawn-extruded rods, for the melt-
Keywords:
Carbon nanotubes
spun fibers 4c was found to be highly dependent on the DDR. Stiffness and offset yield stress of the
Polymer fibers MWCNT/PC fibers were increased starting with the addition of 1 wt% of MWCNTs. However, decreased
Piezoresistivity spinnability was observed for fibers with MWCNT concentrations above 4 wt%. The coefficient that re-
Strain sensing lates the electrical resistance changes to the mechanical strain (strain gage factor SGF) was estimated for
Melt spinning fibers with MWCNTs amounts that range from 3.5 to 6 wt%. The highest SGF ~16 was found at 3.5 wt% and
the lowest SGF ~2.5 at 6 wt% MWCNT. Smart multifunctional textiles, made from such conductive fibers,
have wide applications such as structural health monitoring, detection of gases and liquids, sensor arrays
and flexible sensors.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Current strain sensing technology

The fast growth of technological needs has led to the develop- Specifically in the case of strain sensing and structural health
ment of new functional materials with enhanced properties such as monitoring there are already some commercial technologies
sensing capabilities. It is well known that the addition of multiwall available. A short summary of some of them will be given next.
carbon nanotubes (MWCNTs) improves electrical and mechanical Metallic/semi-conductive strain-gauges (MSGs) are probably the
properties of polymeric materials [1e3]. For this reason CNTs have most widely used sensors. However, due to their morphology and
been extensively studied for the fabrication of nanocomposites shape, placing the sensor is a key factor which in complex struc-
with several potential applications [4e6]. Furthermore, the addi- tures would require a lot of interconnected MSG arrays with a
tion of MWCNTs has been proven to provide polymers with sensing tendency for experimental results to be misinterpreted. Other types
capabilities [7,8]. The versatility of CNT/polymer composite sensors of mechanical strain measuring devices are extensometers. They
has been reflected recently by its multiple uses for sensing liquids are very accurate and reliable, although mounting them in complex
[9], gases [10], chemicals [11], pressure [12], force [13], strain as structures is impossible in practice; this technology is more suitable
well as for structural health monitoring [14]. for laboratory conditions. Using strain sensors made of magnetic
film deposited on copper wires (MFCW) is a good approach because
it can be easily placed onto nearly any structure. However, it re-
quires a secure and direct electrical connection to provide a mag-
netic excitation. Fiber optic sensors have the same advantages as
* Corresponding author.
€tschke).
E-mail address: poe@ipfdd.de (P. Po MFCW. Most of them work using the fiber Bragg grating principle

http://dx.doi.org/10.1016/j.polymer.2015.11.030
0032-3861/© 2015 Elsevier Ltd. All rights reserved.
182 J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189

which is very accurate. This kind of sensor however, has the same natural rubber substrate. These polymer composite based sensors
limitation as MFCW and also is highly sensitive to transverse strains seem to be very favorable for measuring strain, similar to some of
and temperature. Another type of strain sensor available is the the commercial strain sensors which have the disadvantage of not
piezoelectric transducer embedded in miniaturized integrated cir- being able to be freely placed in complex assemblies or in long
cuits. Even when placing them in complex structures they are structures such as beams, pipes, columns, etc.
reliable and robust. Unfortunately, this type of sensor demands
expensive and complex manufacturing processes, as well as so- 1.3. Non-polymer fibers and their composites for strain sensing
phisticated instrumentation and data acquisition to interpret the
electrical signals. In addition to all these drawbacks, the cost of all Even when the previously described composites seem to be a
the above described devices is high, and their performance is still good alternative for strain sensor fabrication, fibers have the
not optimal. advantage of being more flexible than bulk materials, as well as
they can be woven together with other fibers for smart textile
1.2. Polymer based strain sensors fabrication. Additionally, they can be fabricated as very long fibers.
Thanks to these advantages the fibers are easier to place in any kind
An interesting approach for strain sensing applications is the use of structure with all kind of shapes and lengths. Hence, fibers with
of composite materials filled with carbon fillers. Due to the excel- piezoresistive properties offer a very promising approach for sensor
lent viscoelastic properties of polymers and their processing fabrication. Nevertheless, as is the case for bulk materials the
versatility, carbon particle filled polymers have been widely studied processing technique and conditions are key factors for the final
for their use as strain sensors. It is possible to know the strain properties of the fiber sensor. Several studies have reported the
sensitivity of composite materials by quantifying the changes in performance of fibers as strain sensors, though the majority of
their electrical resistance upon a mechanical stimulus (e.g. tension published work is based on non-polymer fibers.
or compression), a phenomenon known as piezoresistivity. The By using carbon filaments (0.15 mm of diameter and length
sensitivity magnitude is commonly referred as the gauge factor >100 mm) embedded into polyether sulfone, Shui et al. were able to
(SGF) and it is the quotient between the relative resistance change accurately measure strain below 1% under cyclic loadings [26].
(DRrel ¼ (RactualRinitial)/Rinitial in %) and the applied mechanical Embedded fiber optic sensors into concrete beams subjected to
strain (Dε in %). Murugaraj et al. reported a maximum SGF of ~7 on tension and flexure has been reported by Belarbi et al. finding that
polyimide films filled with 1.5e8 vol% carbon nanoparticles upon the sensor can keep working even beyond the beam break [27]. For
tensile stress [15]. Luheng et al. found a direct relationship between a hybrid material made from carbon fiber (CF)/CNT/epoxy com-
the carbon black (CB) weight concentration and the piezoresistivity posite Lee et al. found that the composite had a better sensing
of silicon rubber composites which is related to the number of performance when adding CNTs than when using CF alone [28].
effective conductive paths [16]. Kim et al. evaluated piezoresistivity More recently Boehle et al. made a strain sensor for structural
on epoxy/graphene composites finding SGF as high as 11.4 within health monitoring by using glass fiber (GF) with CNTs and carbon
the range of 1000 micro-strain (i.e. the strain producing a defor- nanofibers grown on the GF surface finding that the fiber can follow
mation of 1000 parts per million) [17]. Polymer/MWCNT compos- the true mechanical strain and has a SGF similar to that of
ites have been also widely evaluated as strain sensors showing to be commercially available MSGs [29]. Similar findings were obtained
a very promising alternative for future commercial applications by Gao et al. for GF coated with MWCNT by dip-coating [30].
[18]. Nevertheless, the kind of polymer, shape, morphology and
design of the CNT composite play also a very important role in 1.4. Polymer fibers for strain sensing
strain sensing and structural health monitoring materials devel-
opment. Dog-bone samples made from CNT reinforced thermosets All the previously described works show the great effort to
like vinyl ester resins have proven to provide very accurate sensing achieve accurate and reliable sensor materials consisting either of
and detection of the elastic and plastic regimes upon compression glass or carbon fibers or polymer composites made there from.
[19]. However, the electrical response shows a nonlinear and non- However, polymeric fibers can be regarded are much more versatile
monotonic behavior which makes it difficult to transfer it to prac- as their properties can be fine-tuned by the choice of conductive
tical applications. Rein et al. proposed a sensor made from MWCNT fillers, amounts and filler combinations and their geometries. The
buckypaper (BP) for measuring strain onto an epoxy dog-bone published work on the use of polymer fibers as strain sensors is
sample [20]. The BP sensor was shown to be very accurate and rare. In a study by Denis-Lutard et al., MWCNT/polyvinyl alcohol
sensitive, yet high amount of nanotubes are needed to fabricate a wet-spun fibers exhibited an approximate strain sensitivity of 1.5
BP. A multilayer polyelectrolyte/carbon nanotube composite of Ucm/% of strain [31]. This illustrates that polymer/MWCNT fibers
single-walled CNTs (SWCNT) prepared by a layer-by-layer assembly are good candidates for strain sensing applications. In studies by
method onto a polyvinyl alcohol substrate has been evaluated by Mattmann et al. filament fibers made from styreneeethylene/
Loh et al. [21]. They found that increasing the CNT concentration butyleneestyrene (SEBS)/CB could be used to successfully measure
used to form each layer as well as increasing the number of layers in-situ the deformation of a textile [32]. However, 50 wt% of filler
improves the sensing properties of the composite. On the other was needed to achieve suitably high conductivity values in the
hand, elastomer polyurethane/MWCNT films have the advantage to SEBS/CB filaments for accurate sensing. These filaments were
withstand and sense strains as large as 400% [22,23]. Subramani fabricated using a capillary rheometer and had thicknesses in the
et al. proposed the use of a flexible strain sensor made of poly- range of 300e315 mm. Interestingly, Melnykowycz et al. found that
chloroprene rubber with an ionic liquid (IL)-modified MWCNTs the performance of self-prepared SEBS/CB monofilaments is better
that combines the advantages of high stretchability of the rubber than that of a commercial filament Merlin® made of carbon
with high conductivity of the IL-MWCNTs [24]. More recently, impregnated rubber. They found that melt-spinning the polymer
Tadakaluru et al. proposed a strain sensor made of MWCNT/ together with CB enables higher sensing capability achieving SGF up
graphite powder deposited onto the surface of a pre-stretched to 19 for fibers with 50 wt% of CB as compared to the SGF of com-
natural rubber film [25]. They achieved very accurate reproduc- mercial strain gauges which is around 2 [33,34].
ibility. Over 400 loading cycles for a strain range of 150%e500% and At present, commercially available strain sensor fibers are very
a maximum strain measurement capacity of 620% limited by the limited and commercial melt-spun fibers as strain sensors are not
J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189 183

yet in the market. On the other hand, melt spinning has been
shown to be efficient suitable technique for the fabrication of
conductive fibers made from polymers filled with MWCNTs
[35e39]. It is also known that a lower amount of MWCNT is needed
to achieve the same or higher conductivities as compared to other
traditional conductive fillers, like CB. Having polymer/MWCNT fi-
bers with sensing capabilities enables the fabrication of smart
multifunctional textiles. Through their versatility, smart multi-
functional textiles have wide applications such as strain sensing,
structural health monitoring, detection of gases and liquids, flexible
sensors and many more. However, the properties (e.g. strain
sensing, electrical resistivity, etc.) and behavior of MWCNT/poly-
mer composite fibers are not completely understood and further
research is needed.
In this work, commercially available MWCNTs were incorpo-
rated into polycarbonate (PC) by melt mixing, and subsequently,
MWCNT/PC fibers were fabricated by melt-spinning in a piston-
type spinning device by varying the spinning conditions, espe-
cially the draw down ratio. Electrical, mechanical, and strain
sensing (piezoresistivity) properties of these fibers were evaluated
as a function of the MWCNT concentration (1e6 wt%) and spinning
conditions.

2. Experimental

2.1. Materials and fiber fabrication

Fig. 1. Schema of the piston type spinning device, constructed at IPF.


Commercially available MWCNTs (Nanocyl™ NC7000; Nanocyl
S.A., Belgium) as powder and polycarbonate (Makrolon® 2205;
Bayer MaterialScience AG, Germany) were melt-mixed in a DSM 2.2. Characterization of the state of nanotube dispersion
microcompounder at 280  C for 5 min using a screw speed of
250 rpm. Dried PC pellets (100  C, 8 h) and MWCNT powder were Light microscopy photographs were taken in transmission mode
manually mixed and then added to the running compounder. The using a BH2 microscope combined with a DP71 camera (Olympus
MWCNT weight concentration was varied from 1 to 6 wt%. Prior to Deutschland GmbH, Germany) of the melt-mixed MWCNT/PC
spinning, the pelletized composite material obtained from extruded strands for investigating the state of nanotube dispersion
microcompounding was dried again in the vacuum oven at 100  C before melt-spinning. For this, thin cuts with a thickness of 10 mm
for 8 h. Melt-spinning was carried on a piston-type spinning de- were prepared using a glass knife perpendicular to the extrusion
vice (Fig. 1) constructed at IPF with a spinneret die hole diameter direction. From each of the images, the agglomerate area ratio (A/
of 0.6 mm. A0) was determined. This is the ratio of the cumulative area of the
A melt temperature of 280  C was used and the pellets were pre- projected MWCNT agglomerates (A) to the total area of the
heated for 5 min in the barrel before spinning. Different through- micrograph (A0) and was calculated using the digital image pro-
puts (0.47e1.17 cm3/min) and take-up velocities (20e800 m/min) cessing software ImageJ on Analyze Particles mode. A minimum of
were used and selected according to the homogeneity obtained on 10 thin cuts per sample was used for this dispersion analysis.
the fibers [36]. By varying the throughput and take-up velocity
different draw dawn ratios were obtained. The draw dawn ratio
(DDR) can be defined as [40], 2.3. Resistivity measurements

DDR ¼ vL =v0 (1) In order to evaluate the effect of the processing conditions
during spinning on electrical resistivity (r) of the MWCNT/PC
(where vL ¼ velocity of the fiber at the winder and v0 ¼ velocity at
composites, r was measured on melt-mixed compression molded
the die).
samples and on melt-spun fibers. The pelletized extruded strands
According to the law of conservation of mass and the continuity
from melt-mixing were compression molded at 280  C for 2 min
equation the throughput can be expressed as,
into circular plates of 60 mm diameter and 0.5 mm thickness using
 .  a Weber hot press (Model PW EH, Paul Otto Weber GmbH, Ger-
Throughput ¼ p 4D20 *v0 (2) many). Electrical resistivity was measured using a Keithley elec-
trometer 6517A combined with a Keithley 8009 test fixture for
(where D0 ¼ diameter of the die hole). resistances higher than 107 U on disc shaped samples (triangles in
Thus, the DDR can be calculated by using the following equation, Fig. 4). For resistances lower than 107 U strips with the dimensions
30  3  0.5 mm3 were measured using a four-point probe (squares
DDR ¼ ðtake  up speed=throughputÞ*Ac (3) in Fig. 4) [41].
In order to have a first approach on the r of the fibers, in a first
(where Ac ¼ cross sectional area). set the electrical resistivity was measured on composites contain-
For further discussions only the DDR will be referred hereafter. ing 1 wt% of MWCNT with DDRs varying between 30 and 480 and
184 J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189

free sample length was 35 mm (See Fig. 2). The offset yield stress
(soff) at 0.8% (namely, the stress at a strain offset of 0.8% in the
stressestrain curve) was obtained to evaluate the effect of the
addition of MWCNTs on the plastic yielding of the fibers.
Piezoresistivity tests were carried by measuring the change in
electrical resistance DRrel in-situ during tensile tests until sample
break. The same test conditions and specimen dimensions were
employed as for the mechanical characterization testing. The
changes in electrical resistance were measured using a Keithley
DMM 2001 electrometer applying software for data acquisition
with a sampling rate of 1s developed at IPF. The fibers were con-
tacted using electrically conducting silver paint and connected to
the measuring instrument with copper wires, as illustrated in Fig. 2.
Fig. 2. Schema of the measuring set-up for piezoresistive testing of fibers.

3. Results and discussion


compared to undrawn fibers and the compression molded samples.
3.1. State of nanotube dispersion in the composites
Based on these results, a second set of fibers with more suitable
electrical properties and good spinnability produced at much lower
The spinnability of composites is dependent on the homoge-
DDRs was fabricated and used for further characterizations; these
neity of the material, as agglomerates above a certain size may
DDRs were 4.83, 6.02, 8.08 and 12.03. Electrical resistivity mea-
cause breakage during the spinning process. Thus, the state of
surements of the fibers were carried out on single PC/MWCNT fil-
dispersion of the melt-mixed composites was analyzed before
aments with diameters ranging from ~15 mm (DDR ¼ 480) to
spinning using transmission light microscopy, as shown in Fig. 3.
~250 mm (DDR ¼ 4.83). Electrical resistivity of the fibers was
For the 1 wt% MWCNT composite no larger agglomerates are seen.
measured using a Keithley 8002A test fixture for 60 mm long fibers
The larger agglomerates start from 3.5 wt% MWCNT and become
(circles in Fig. 4).
larger as the wt% of CNTs increases. In order to quantify the quality
of dispersion in the MWCNT/PC composites, the agglomerate area
2.4. Mechanical and piezoresistive characterization of melt spun ratio (A/A0) was determined. All composites with amounts lower
fibers than 3.5 wt% (also 2 and 3 wt%, not shown here) have similar
dispersion state as 1 wt%. The composite with 1 wt% shows a value
Tensile tests on the fibers were carried out with a Zwick-Roell of A/A0 of 0.24% which increases to 0.75% for 3.5 wt% and 3.28% for
universal testing machine and at least 5 specimens were evalu- composites with 6 wt% of MWCNTs.
ated for each set of samples. The load cell, the test speed and This shows that good dispersion was successfully achieved for
sample's length were 20 N, 5 mm/min and 95 mm, respectively. The composites with low wt%. However, for composites with higher

Fig. 3. Light microscopy of PC/MWCNT composites after melt-mixing (compression molded plates) for three different MWCNT weight concentrations together with the calculated
agglomerate area ratio A/A0.
J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189 185

MWCNT/PC (disc) reason, a second set of fibers with MWCNT contents varying from 1
18
MWCNT/PC (strip) to 6 wt% and DDRs lower than 30 was fabricated to identify DDRs
16 MWCNT/PC as extruded (DDR = 1) with suitable electrical resistivity for piezoresistive measurements
Log(ρ) (ρ in Ω cm)

MWCNT/PC fiber (DDR = 30) (Fig. 5).


14 MWCNT/PC fiber (DDR = 120) As shown in Fig. 5, even for DDRs as low as 4.83 and 12.03 there
12
MWCNT/PC fiber (DDR = 480) is an evident difference between the r of the corresponding fibers.
For instance, r of fibers with 2 wt% MWCNTs is 4 orders of
10 magnitude lower for DDR ¼ 4.83 than for DDR ¼ 12.03. After
increasing the amount of MWCNTs the difference in r decreases
8
though it is still visible until 5 wt%. It is worth mentioning that fi-
6 bers with 6 wt% of MWCNTs and higher DDR than 4.83 could not be
produced, as fiber breakage occurred preventing the generation of
4 continuous fibers. This may be related to the size of the
2
agglomerates.

0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
12
MWCNT concentration (wt%)
11
Fig. 4. Electrical resistivity of PC/MWCNT composites as a function of the MWCNT
concentration, sample set 1. 10
DDR = 12.03

Log (ρ) (ρ in Ω cm)


9 DDR = 8.08
nanotube amounts there are still primary agglomerates remaining
8 DDR = 6.02
after melt-mixing. This indicates that melt-spinning of composites DDR = 4.83
with higher MWCNT amounts may be more difficult, due to the 7
existence of agglomerates. It is expected that the melt-spinning
6
shaping step does not contribute to further nanotube dispersion
and that such agglomerates are also existent within the melt-spun 5
fibers.
4
3.2. Electrical resistivity of pressed plates and melt-spun fibers 3 ρ suitable for sensing

When studying the strain sensing performance of conductive


2
composites, a certain low resistivity value r (about 50 MU cm) has 1
to be given in the sample before stretching. During stretching, in- 0 1 2 3 4 5 6
crease in resistance is expected which needs to be followed with MWCNTconcentration (wt%)
the electrometers applied. Thus, the electrical resistivity of the
Fig. 5. Electrical resistivity of MWCNT/PC fibers at different DDRs as a function of
initial samples was investigated in their different shapes, as shown
MWCNT concentration, sample set 2.
in Fig. 4 for the PC/MWCNT samples from set 1.
In the case of compression molded samples, a significant
decrease in r was found after the addition of 0.25 wt% MWCNT and 3.3. Mechanical properties of melt spun fibers
percolation occurs below 0.5 wt%. At the addition of 0.5 wt%, r
decreases by 13 orders of magnitude. On the other hand, melt-spun The characteristic stressestrain curves of the PC/MWCNT fibers
fibers have significantly higher r depending on the DDR, as shown for the evaluated DDRs are shown in Fig. 6. For easier comparison,
for the 1 wt% samples. Even the extrudate taken from the spinning the scales are the same for all the graphs shown. Table 1 lists the
device without winding (DDR ¼ 1) has higher r than the tensile properties of PC/MWCNT fibers with a DDR of 4.83. It should
compression molded sample. This suggests that the shaping pro- be mentioned that the other DDRs evaluated (not shown in Table 1)
cess of the composite material significantly influences the exhibited very similar trends to those shown in Table 1 for
morphology of the conducting MWCNT network and by that the DDR ¼ 4.83. It is evident from Fig. 6 and Table 1 that the addition of
measured resistivity and the percolation threshold 4c of the MWCNTs has a strengthening effect on the polymer fibers. An in-
structures. crease in the tensile strength (smax) is visible from the addition of
Consequently, 4c of the melt-mixed and compression molded 2 wt% of MWCNTs reaching a maximum of ~30% increase in fibers
samples is below 0.5. For melt-spun fibers, 4c shows a dependency with 6 wt%. Similarly, as seen from the magnitudes of the elastic
on the DDR, as different MWCNT network structures are formed modulus (E) a stiffening effect was achieved as from the addition of
depending on DDR. The main effect is an expected and confirmed 1 wt% reaching a reinforcement of ~50% for 6 wt% MWCNT/PC fi-
orientation of individualized MWCNTs and smaller agglomerates bers. However, it is also seen that the addition of MWCNTs reduces
while the composite is pushed through the spinneret hole. As the strain at break (εmax) of the fibers which is known to be typical
shown in TEM images of PC-MWCNT fibers, also prepared at for particle filled polymers. Ductility of the fibers was decreased
different take-up velocities [35], this is likely to happen along the upon increased MWCNT content as well as for lower DDRs. The
fiber axis on melt-spun fibers. Interestingly, the difference in r of decrease in ductility with DDR was visible even for the neat PC fiber
melt-spun fibers at DDR ¼ 30 and at DDR ¼ 480 is of only 2 orders where εmax was reduced from ~80% (DDR ¼ 12.03) to <40%. This
of magnitude while the difference in r of fibers at DDR ¼ 30 and reduction in εmax was more pronounced after the addition of
DDR ¼ 1 is of approximately 6 orders of magnitude. Additionally, MWCNTs. Interestingly, when comparing the tensile properties of
the r of fibers is already at the DDR of 30 higher than 109 U cm and the fibers with different DDRs at the same amount of MWCNTs,
therefore not suitable for strain sensing applications. For this
186 J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189

80 DDR = 4.83 (ø = 260 ± 8 μm) DDR = 6.02 (ø = 234 ± 7 μm)


70
60

Stress (MPa)
50
PC neat PC neat
40 1% MWCNT/PC 1% MWCNT/PC
2% MWCNT/PC 2% MWCNT/PC
30
3% MWCNT/PC 3% MWCNT/PC
20 4% MWCNT/PC 4% MWCNT/PC
10 5% MWCNT/PC 5% MWCNT/PC
6% MWCNT/PC 6% Not spinnable
0
80 DDR = 8.08 (ø = 200 ± 5 μm) DDR = 12.03 (ø = 164 ± 7 μm)
70
60
Stress (MPa)

50
PC neat PC neat
40 1% MWCNT/PC 1% MWCNT/PC
30 2% MWCNT/PC 2% MWCNT/PC
3% MWCNT/PC 3% MWCNT/PC
20 4% MWCNT/PC 4% MWCNT/PC
10 5% Not spinnable 5% Not spinnable
6% Not spinnable 6% Not spinnable
0
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Strain (%) Strain (%)

Fig. 6. Stress-strain curves of the PC/MWCNT fibers prepared at different DDRs; average diameter (ø) of the fibers are also shown.

Table 1
Tensile properties of PC/MWCNT fibers prepared at DDR ¼ 4.83.

MWCNT concentration (wt%)

0 1 2 3 4 5 6

E (GPa) 2.25 ± 0.02 2.35 ± 0.08 2.56 ± 0.04 2.58 ± 0.05 2.72 ± 0.04 2.98 ± 0.14 3.33 ± 0.17
smax (MPa) 60.3 ± 0.54 60.0 ± 1.48 64.8 ± 0.77 66.3 ± 1.07 69.4 ± 1.46 73.9 ± 6.31 77.2 ± 6.27
soff (MPa) 47.0 ± 0.69 50.3 ± 0.84 53.4 ± 1.27 53.6 ± 1.79 53.4 ± 1.24 57.6 ± 3.04 59.1 ± 2.64
εmax (%) 32.7 ± 11.0 24.1 ± 2.81 19.6 ± 4.25 28.7 ± 7.67 19.0 ± 2.69 6.88 ± 1.47 8.30 ± 0.84

mechanical reinforcement was more significant for fibers made more pronounced at larger DDRs where the size of the primary
with lower DDR. agglomerates becomes more significant due to the reduced cross-
As can be seen from Table 1, an increase in soff of the fibers is section of the fibers. At 6 wt% MWCNT loading, the largest ag-
visible for all the wt% evaluated where a maximum of ~25% increase glomerates seen in Fig. 1 are in the same size as the fiber diameters
was achieved for fibers with DDR ¼ 4.83 at 6 wt% MWCNTs. This of 164e260 mm. Thus, above DDR ¼ 6.02 the composite was not
implies that the addition of MWCNTs has an important effect on spinnable anymore, which starts for samples with 5 wt% loading at
increasing the stress required to reach plastic deformation of the DDR ¼ 8.08. This indicates that improving dispersion could be a
fibers. These results are in accordance with Zetina-Herna ndez et al. major need to get better performing fibers. These findings are in
who also found an effect on the polymer yielding after the addition accordance with studies by Po €tschke et al. who found that the
of 4e10 wt% of MWCNTs in compression molded polypropylene addition of MWCNT into polylactid acid (PLA) in amounts above
dog-bone samples [42]. This phenomenon can be explained by the percolation threshold leads to a significant reduction of flow
wrapping of polymer chains around MWCNTs. These polymer- properties of the melt due to increased fillerefiller interaction [36].
MWCNT entanglements restrict the movement of the polymer They also found that the addition of 5 wt% of MWCNT into PLA
chains while they are subjected to tensile loading. Similarly, when resulted in inhomogeneous fiber diameters which led to fiber
the MWCNT amount increases, the energy needed to overcome the breakage.
physical interactions (mainly van der Waals forces) between a
larger amount of MWCNT and polymer chains to reach the plastic
deformation during tensile test also increases. Likewise, the strain 3.4. Piezoresistive properties of melt spun fibers
at break is also affected since after the addition of MWCNTs the
strain at which breakage occurs is significantly decreased as a result Piezoresistive properties of MWCNT/PC fibers with r suitable for
of the reduced mobility of the polymer chains. sensing applications are presented in Fig. 7. The electrical response
It is also seen in Fig. 6 that reduced spinnability was found with of the piezoresistive samples during tensile tests is shown for PC/
increasing MWCNT amount which also may be a consequence of MWNCT fibers produced at 4 different DDRs with MWCNT loadings
the reduced mobility of the polymer chains upon addition of ranging from 3 to 6 wt%. The tests were carried out until sample
MWCNTs. In addition, the higher number of remaining primary break.
nanotube agglomerates seen after melt-mixing (Fig. 1) may also As discussed in Section 3.2, not all the fibers were suitable for
contribute to breakage at lower strains. The fiber breakage was the measurements of resistance changes. Additionally as discussed
in Sections 3.1 and 3.3, fibers could not be produced at high wt%
J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189 187

DDR = 4.83 DDR = 6.02


80
3% MWCNT/PC
60 3.5% MWCNT/PC 3% ρ was too high
ΔR/R 4% MWCNT/PC 3.5% MWCNT/PC
(%)
5% MWCNT/PC 4% MWCNT/PC
40 6% MWCNT/PC 5% MWCNT/PC
6% Not spinnable

20

0
DDR = 8.08 DDR = 12.03
80
3% ρ was too high
3.5% ρ was too high
60
3% ρ was too high 4% MWCNT/PC
ΔR/R0
3.5% MWCNT/PC 5% Not spinnable
(%) 6% Not spinnable
40 4% MWCNT/PC
5% Not spinnable
6% Not spinnable
20

0
0 2 4 6 8 0 2 4 6 8
Strain (%) Strain (%)

Fig. 7. Electrical response of PC/MWCNT fibers of sample set 2 during tensile testing for different DDRs. The vertical line in the plot of DDR ¼ 4.83 illustrates the strain limit of 1.8%
used for the calculation of SGF.

due to reduced spinnability resulting from remaining MWCNTs composite films is less sensitive to the internal density and size of
primary agglomerates. Therefore, depending on their spinnability CNT agglomerates than the electrical conductivity [47].
and the achieved resistivity values, the piezoresistivity was only In order to quantify the strain sensitivity of the different
characterized for fibers with 3 wt% and higher. In the case of the MWCNT/PC fibers, the gage factor (SGF) was calculated and is
highest DDR (12.03) only one weight concentration (4 wt%) could shown in Fig. 8. The SGF was estimated as the slope in the range
be measured. In order to compare the strain sensing capabilities at from 0 to 1.8% strain from the characteristic DR/R0 vs. strain curves
low wt%, a composite with a MWCNT amount of 3.5 wt% was (see vertical line on Fig. 7, DDR ¼ 4.83). As shown in Fig. 8, fibers
additionally prepared and studied. with higher DDRs have higher SGF and sensitivity shows a larger
From Fig. 7 it is visible that the electrical response increases decrease with the nanotube amount; while the fibers with the
upon increasing strain. For instance, in average for fibers with lowest DDR have significantly lower SGF values which show a much
3.5 wt% MWCNTs and DDR ¼ 4.83 the electrical response continues smaller decrease with nanotube amount. For instance, fibers with
increasing until the samples break at about ~5.5% of strain. For fi- 3.5 wt% of MWCNT that were melt-spun at a DDR ¼ 8.08 have a SGF
bers with 4 wt% and 5 wt% (excepting DDR ¼ 8.08) the electrical three times larger (~16) than fibers with 4 wt% at the same DDR
response reaches a maximum and then decreases slightly until
sample break which occurs between 4 and 5% strain depending on
the DDR. This behavior is due to the necking effect in the fibers 18
during stretching. Nevertheless, for fibers with 3.5 wt% of MWCNT
and DDR ¼ 6.02 the change in electrical resistance is as high as 80% 16
at 6% of strain, which is indicative of high sensitivity. Furthermore, DDR = 12.03
14 DDR = 8.08
the difference in the strain sensitivity of fibers at DDR ¼ 4.83 with
DDR = 6.02
Gage Factor SGF(-)

3 wt% and 3.5 wt% is significant. There are several theories for the 12
working mechanisms in the piezoresistive nanocomposite strain DDR = 4.83
sensors. Among these, the most accepted are: (a) significant 10
changes in the conductive network structure formed by CNTs [43],
8
(b) changes in tunneling resistance between neighboring CNTs due
to distance change [44] and (c) piezoresistivity of CNTs themselves 6
due to their deformation [45]. For our case the first two are the
most relevant and likely to happen. This can be concluded from the 4
findings that melt-spinning promotes MWCNT orientation along
the fiber axis [35] and that it has been stated that the CNT
2
piezoresistive sensitivity has as a very small contribution (~5%) of 0
the whole piezoresistive effect of the composite [46]. It may be also 3 4 5 6
assumed that the internal density and size of the CNT agglomerates MWCNT concentration ( wt%)
have an effect on the piezoresistive properties of the fibers. How-
ever, Gong et al. found that the piezoresistivity of the CNT polymer Fig. 8. Gage factors SGF of the PC/MWCNT fibers at different DDRs as a function of the
MWCNT concentration.
188 J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189

(~5). This indicates that the SGF is highly sensitive to the weight resistivity of the fibers. Piezoresistivity effects could be character-
concentration of the MWCNT and to the DDR employed to produce ized only for fibers with MWCNT amounts of 3 wt% or higher. In the
the fibers. case of the highest DDR (12.03) only one weight concentration
This sensitivity of SGF on MWCNT wt% and DDR can be explained (4 wt%) could be evaluated. SGF was found to be very sensitive to the
when considering the MWCNT network structure within the fibers. amount of MWCNTs and to the DDR. Lower MWCNT wt% and
The change in electrical resistance upon elongation gets higher as higher DDR led to higher SGF. The highest SGF found in this study
the nanotube concentration is closer to the electrical percolation was ~16 for fibers with 3.5 wt% MWCNT prepared at a DDR of 8.08.
threshold concentration. The change is much lower if the MWCNT This value is promisingly high compared to commercial available
amount is well above 4c. This is due to the fact that a less dense but MSG having values in the range of 2.
already existing conducting network inside the polymer is For this kind of polymer based fiber strain sensors there is a
deformed easily and some of the nanotubes are more easily wide list of potential applications ranging from health monitoring
disconnected upon external stimuli and therefore the change in to robotics.
electrical resistance is much larger as compared to composites with
higher density of conductive elements. In the latter, external Acknowledgments
stimuli do not have a considerable effect on the density of
conductive pathways and, thus, on the electrical resistance. Similar Author Bautista-Quijano wants to thank to The National Council
behavior was also found by Villmow et al. who studied liquid of Science and Technology of Mexico (CONACYT) and The National
sensing sensitivity on polycarbonate/MWCNT compression molded n State
Council of Science, Innovation and Technology of the Yucata
composites and found higher sensitivity in conductive composites (CONCIYTEY) for the financial support.
with lower MWCNT amounts [48].
When drawing fibers the used DDR also modifies the perco- References
lating network structure inside the polymer. As the DDR increases,
the nanotube orientation along the fiber increases leading to a less [1] E.T. Thostenson, C. Li, T.-W. Chou, Nanocomposites in context, Compos. Sci.
dense network with more interrupted pathways as compared to a Technol. 65 (3e4) (2005) 491e516.
[2] M. Terrones, Science and technology of the twenty-first century: synthesis,
nondrawn sample. Thus, 4c increases with DDR. At an actual CNT properties, and applications of carbon nanotubes, Ann. Rev. Mater Res. 33 (1)
concentration, the difference to 4c gets smaller with increasing (2003) 419e501.
DDR. [3] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Carbon Nanotubesethe Route
toward applications, Science 297 (5582) (2002) 787e792.
Therefore, there is compromise between the strain sensitivity,
[4] M.F.L. De Volder, S.H. Tawfick, R.H. Baughman, A.J. Hart, Carbon nanotubes:
the MWCNT wt% and the DDR. The compromise is that the re- present and future commercial applications, Science 339 (6119) (2013)
sistivity has to be in a stable measurable range, and at the same 535e539.
time high resistive samples always result in larger sensitivity. [5] J. Robertson, Realistic applications of CNTs, Mater. Today 7 (10) (2004) 46e52.
[6] V.N. Popov, Carbon nanotubes: properties and application, Mater. Sci. Eng. R
It is worth mentioning that even the lowest SGF found in this Rep. 43 (3) (2004) 61e102.
work (~2.5 for 6 wt% and DDR ¼ 4.83) is above the SGF of conven- [7] C. Gao, Z. Guo, J.-H. Liu, X.-J. Huang, The new age of carbon nanotubes: an
tional MSGs which generally is close to 2 [34]. The SGF's found in updated review of functionalized carbon nanotubes in electrochemical sen-
sors, Nanoscale 4 (6) (2012) 1948e1963.
this work are also higher than the ones found in literature for dog- [8] C. Li, E.T. Thostenson, T.-W. Chou, Sensors and actuators based on carbon
bone shaped thermoset samples and compression molded or so- nanotubes and their composites: a review, Compos. Sci. Technol. 68 (6) (2008)
lution cast thermoplastic films [49e51]. For other polymer fibers, 1227e1249.
[9] T. Villmow, S. Pegel, A. John, R. Rentenberger, P. Po € tschke, Liquid sensing:
similar SGF were only found after filling with ~50 wt% of CB [32,33] smart polymer/CNT composites, Mater. Today 14 (7e8) (2011) 340e345.
which is much higher than the MWCNT contents used in this study. [10] Q. Fan, Z. Qin, T. Villmow, J. Pionteck, P. Po€ tschke, Y. Wu, et al., Vapor sensing
properties of thermoplastic polyurethane multifilament covered with carbon
nanotube networks, Sensors Actuators B Chem. 156 (1) (2011) 63e70.
4. Conclusions
[11] L. Han, A.L. Andrady, D.S. Ensor, Chemical sensing using electrospun polymer/
carbon nanotube composite nanofibers with printed-on electrodes, Sensors
Fibers made of PC with different MWCNT weight concentrations Actuators B Chem. 186 (0) (2013) 52e55.
[12] H.-M. So, J.W. Sim, J. Kwon, J. Yun, S. Baik, W.S. Chang, Carbon nanotube based
were fabricated by melt-mixing and subsequent melt-spinning. It
pressure sensor for flexible electronics, Mater. Res. Bull. 48 (12) (2013)
was found that MWCNT are dispersed well in PC for composites 5036e5039.
with low wt%. Higher amounts than 3 wt% resulted in the existence [13] M. Vatani, E.D. Engeberg, J.-W. Choi, Force and slip detection with direct-write
of residual MWCNT agglomerates. MWCNT/PC fibers were suc- compliant tactile sensors using multi-walled carbon nanotube/polymer
composites, Sensors Actuators A Phys. 195 (0) (2013) 90e97.
cessfully melt-spun by varying the draw down ratio DDR by [14] N. Hu, Y. Karube, M. Arai, T. Watanabe, C. Yan, Y. Li, et al., Investigation on
changing throughput and take-up velocity. However, the presence sensitivity of a polymer/carbon nanotube composite strain sensor, Carbon 48
of agglomerates in the composite material starting at 3 wt% (3) (2010) 680e687.
[15] P. Murugaraj, D.E. Mainwaring, N. Mora-Huertas, Electromechanical response
MWCNTs significantly reduced the capability of the composites to of semiconducting carbonepolyimide nanocomposite thin films, Compos. Sci.
be melt-spun. As a consequence, it was not possible to produce Technol. 69 (14) (2009) 2454e2459.
fibers with loadings of MWCNT higher than 5 wt% and DDR values [16] W. Luheng, D. Tianhuai, W. Peng, Influence of carbon black concentration on
piezoresistivity for carbon-black-filled silicone rubber composite, Carbon 47
higher than 4.83. In mechanical testing, an increase in the stiffness (14) (2009) 3151e3157.
of the fibers was seen for all the wt% evaluated, as well as a [17] Y.-J. Kim, J.Y. Cha, H. Ham, H. Huh, D.-S. So, I. Kang, Preparation of piezor-
reduction in the strain at break. A relationship between the offset esistive nano smart hybrid material based on graphene, Curr. Appl. Phys. 11
(1, Suppl.) (2011). S350eS2.
yield stress of the fibers and the amount of MWCNTs was found [18] Hu N. Alamusi, H. Fukunaga, S. Atobe, Y. Liu, J. Li, Piezoresistive strain sensors
suggesting an effect on the polymer yielding due to the addition of made from carbon nanotubes based polymer nanocomposites, Sensors 11 (11)
the MWCNTs. Mechanical reinforcement was more significant for (2011) 10691e10723.
[19] J.J. Ku-Herrera, F. Avile s, G.D. Seidel, Self-sensing of elastic strain, matrix
fibers made with the lower DDR evaluated (4.83).
yielding and plasticity in multiwall carbon nanotube/vinyl ester composites,
Electrical percolation of melt-mixed PC/MWCNT compression Smart Mater. Struct. 22 (8) (2013) 085003.
molded composites occurred below 0.5 wt% while for PC/MWCNT [20] M.D. Rein, O. Breuer, H.D. Wagner, Sensors and sensitivity: carbon nanotube
fibers 4c was shifted to higher values dependent on the DDR used to buckypaper films as strain sensing devices, Compos. Sci. Technol. 71 (3)
(2011) 373e381.
produce the fibers. There is a compromise between spinnability [21] K.J. Loh, J.P. Lynch, B.S. Shim, N.A. Kotov, Tailoring piezoresistive sensitivity of
(depending on state of nanotube dispersion) and suitable electrical multilayer carbon nanotube composite strain sensors, J. Intelligent Material
J.R. Bautista-Quijano et al. / Polymer 82 (2016) 181e189 189

Syst. Struct. 19 (7) (2008) 747e764. melt-spun polyethylene/multiwalled carbon nanotube fiber through forma-
[22] J.R. Bautista-Quijano, F. Avile s, J.V. Cauich-Rodriguez, Sensing of large strain tion of a shish-kebab structure, J. Phys. Chem. B 114 (33) (2010)
using multiwall carbon nanotube/segmented polyurethane composites, 10693e10702.
J. Appl. Polym. Sci. 130 (1) (2013) 375e382. [38] T.D. Fornes, J.W. Baur, Y. Sabba, E.L. Thomas, Morphology and properties of
[23] Q. Fan, Z. Qin, S. Gao, Y. Wu, J. Pionteck, E. Ma €der, et al., The use of a carbon melt-spun polycarbonate fibers containing single- and multi-wall carbon
nanotube layer on a polyurethane multifilament substrate for monitoring nanotubes, Polymer 47 (5) (2006) 1704e1714.
strains as large as 400%, Carbon 50 (11) (2012) 4085e4092. [39] J.K.W. Sandler, S. Pegel, M. Cadek, F. Gojny, M. van Es, J. Lohmar, et al.,
[24] K. Subramaniam, A. Das, G. Heinrich, Development of conducting poly- A comparative study of melt spun polyamide-12 fibres reinforced with carbon
chloroprene rubber using imidazolium based ionic liquid modified multi- nanotubes and nanofibres, Polymer 45 (6) (2004) 2001e2015.
walled carbon nanotubes, Compos. Sci. Technol. 71 (11) (2011) 1441e1449. [40] R. Beyreuther, H. Brünig, R. Vogel, Melt Spinning e Polymer Formation Under
[25] S. Tadakaluru, W. Thongsuwan, P. Singjai, Stretchable and flexible high-strain High Stress Conditions. Polymeric Material Encyclopedia: Synthesis, Proper-
sensors made using carbon nanotubes and graphite films on natural rubber, ties and Applications, CRC Press, 1996, pp. 4061e4074.
Sensors 14 (1) (2014) 868e876. [41] P. Po€tschke, A.R. Bhattacharyya, A. Janke, Morphology and electrical resistivity
[26] S. Xiaoping, D.D.L. Chung, A piezoresistive carbon filament polymer-matrix of melt mixed blends of polyethylene and carbon nanotube filled poly-
composite strain sensor, Smart Mater. Struct. 5 (2) (1996) 243. carbonate, Polymer 44 (26) (2003) 8061e8069.
[27] B. Abdeldjelil, E.W. Steve, K. Chandrashekhara, C. Josh, K. Bethany, Smart fiber- [42] O. Zetina-Herna ndez, S. Duarte-Aranda, A. May-Pat, G. Canche -Escamilla,
reinforced polymer rods featuring improved ductility and health monitoring J. Uribe-Calderon, P.I. Gonzalez-Chi, et al., Coupled electro-mechanical prop-
capabilities, Smart Mater. Struct. 10 (3) (2001) 427. erties of multiwall carbon nanotube/polypropylene composites for strain
[28] S.-I. Lee, D.-J. Yoon, Structural health monitoring for carbon fiber/carbon sensing applications, J. Mater Sci. 48 (21) (2013) 7587e7593, 2013/11/01.
nanotube (CNT)/epoxy composite sensor, Key Eng. Mater. 321e323 (2006) [43] P. Myounggu, K. Hyonny, P.Y. Jeffrey, Strain-dependent electrical resistance of
290e293. multi-walled carbon nanotube/polymer composite films, Nanotechnology 19
[29] M. Boehle, Q. Jiang, L. Li, A. Lagounov, K. Lafdi, Carbon nanotubes grown on (5) (2008) 055705.
glass fiber as a strain sensor for real time structural health monitoring, Int. J. [44] N. Hu, Y. Karube, C. Yan, Z. Masuda, Tunneling effect in a polymer/carbon
Smart Nano Mater. 3 (2) (2012) 162e168, 2012/06/01. nanotube nanocomposite strain sensor, Acta Mater 56 (2008) 2929e2936.
[30] S-l Gao, R.-C. Zhuang, J. Zhang, J.-W. Liu, E. M€ ader, Glass fibers with carbon [45] D. Prasad, L. Zhiling, N. Satish, E.V. Barrera, Nanotube film based on single-wall
nanotube networks as multifunctional sensors, Adv. Funct. Mater. 20 (12) carbon nanotubes for strain sensing, Nanotechnology 15 (3) (2004) 379.
(2010) 1885e1893. [46] s, F. Avile
A.I. Oliva-Avile s, G.D. Seidel, V. Sosa, On the contribution of carbon
[31] V. Denis-Lutard, A. Derre, M. Maugey, C. Mercader, S. Moisan, C. Zakri, et al., nanotube deformation to piezoresistivity of carbon nanotube/polymer com-
New wet spinning process for the continuous production of polymer/carbon posites, Compos. Part B Eng. 47 (0) (2013) 200e206.
nanotubes composite fibers, in: C. Binetruy, F. Boussu (Eds.), Recent Advances [47] S. Gong, Z.H. Zhu, S.A. Meguid, Carbon nanotube agglomeration effect on
in Textile Composites, DEStech Publications, Inc, Lille, France, 2010, pp. piezoresistivity of polymer nanocomposites, Polymer 55 (21) (2014)
167e172. 5488e5499.
[32] C. Mattmann, F. Clemens, G. Tro € ster, Sensor for measuring strain in textile, [48] T. Villmow, S. Pegel, P. Po €tschke, G. Heinrich, Polymer/carbon nanotube
Sensors 8 (6) (2008) 3719e3732. composites for liquid sensing: model for electrical response characteristics,
[33] M. Melnykowycz, B. Koll, D. Scharf, F. Clemens, Comparison of piezoresistive Polymer 52 (10) (2011) 2276e2285.
monofilament polymer sensors, Sensors 14 (1) (2014) 1278e1294. [49] G.T. Pham, Y.-B. Park, Z. Liang, C. Zhang, B. Wang, Processing and modeling of
[34] Hottinger-Baldwin-Messtechnik-GmbH. HBM Strain Gauge Catalog. 2015. conductive thermoplastic/carbon nanotube films for strain sensing, Compos.
[35] P. Po€tschke, H. Brünig, A. Janke, D. Fischer, D. Jehnichen, Orientation of mul- Part B Eng. 39 (1) (2008) 209e216.
tiwalled carbon nanotubes in composites with polycarbonate by melt spin- [50] M.H.G. Wichmann, S.T. Buschhorn, J. Gehrmann, K. Schulte, Piezoresistive
ning, Polymer 46 (23) (2005) 10355e10363. response of epoxy composites with carbon nanoparticles under tensile load,
[36] P. Po€ tschke, T. Andres, T. Villmow, S. Pegel, H. Brünig, K. Kobashi, et al., Liquid Phys. Rev. B 80 (24) (2009) 245437.
sensing properties of fibres prepared by melt spinning from poly(lactic acid) [51] J.R. Bautista-Quijano, F. Aviles, J.O. Aguilar, A. Tapia, Strain sensing capabilities
containing multi-walled carbon nanotubes, Compos. Sci. Technol. 70 (2) of a piezoresistive MWCNT-polysulfone film, Sensors Actuators A Phys. 159
(2010) 343e349. (2) (2010) 135e140.
[37] F. Mai, K. Wang, M. Yao, H. Deng, F. Chen, Q. Fu, Superior reinforcement in

You might also like