REVIEW
XXXX
Piezoresistivity, Strain, and Damage Self-Sensing of
Polymer Composites Filled with Carbon Nanostructures
Francis Avil
es,* Andr
es I. Oliva-Avil
es, and Marco Cen-Puc
Previous and current research on piezoresistivity of polymer composites filled
with carbon nanostructures is reviewed. The review covers the use of the
coupled electro-mechanical response of these materials to self-sense their
strain and damage during mechanical loading. The mechanisms yielding
changes in electrical resistance upon mechanical loading in polymer
composites filled with carbon nanostructures are first discussed. Published
knowledge is then summarized, starting with framework literature on carbon
black and graphite and then moving to more recent research on carbon
nanotubes, exfoliated graphite, and few-layer graphene sheets. Piezoresistive
studies of polymer nanocomposites with aligned carbon fillers are also
reviewed. It is aimed that this review contributes in collecting, organizing,
and summarizing the knowledge, foundations, and state of the art on the
piezoresistive response of polymer composites filled with different carbon
allotropes, providing new perspectives and advancing towards the fast
development of smart self-sensing carbon filled nanocomposites.
1. Introduction
Piezoresistivity is defined as the coupling phenomenon between
a mechanical stimulus (strain, e, or stress, σ) and a change in
electrical resistance (ΔR, where R is the electrical resistance).[1,2]
In contrast to piezoelectricity (which goes both ways, mechanical
stimulus to generation of electrical potential, and conversely),
piezoresistivity causes a change only in electrical resistance (not
in electrical potential) and goes only in the mechanical-
to-electrical direction. The first documented report of piezor-
esistivity (for metal wires) is ascribed to William Thomson (Lord
Kelvin) in 1856.[3] Among the first sensors which used this
principle are probably the strain gages, which became
Dr. F. Avil
es, M. Cen-Puc
Centro de Investigaci
on Científica de Yucatán
Unidad de Materiales
Calle 43 No.130
32 y 34, Col. Chuburná de Hidalgo,
97205 M
erida, Yucatán, Mexico
E-mail: faviles@cicy.mx
Dr. A. I. Oliva-Avil
es
Universidad Anáhuac Mayab
Divisi

on de Ingeniería y Ciencias Exactas
Carretera M
erida-Progreso km 15.5 AP 96-Cordemex, CP 97310
M
erida, Yucatán, Mexico
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adem.201701159.
DOI: 10.1002/adem.201701159
Adv. Eng. Mater. 2018, 1701159 1701159 (1 of 23)
www.aem-journal.com
commercial about 75 years after Thom-
son’s discovery. Strain gages are typically
made of thin metal or semiconducting foils
mounted on a backing polymeric film
(typically a polyimide), which are bonded
onto a surface to measure the strain of a
material or structure.[2] Several decades
later, the same basic principle (change in
electrical resistance as a response of an
external stimulus) was implemented to
monitor strain, temperature, moisture, and
chemical vapors (among others) using
polymer composites filled with conductive
fillers such as carbon fibers,[4,5] graph-
ite,[6,7] and carbon black.[8,9] With the rise of
nanomaterials and nanotechnology, this
concept has evolved to the use of carbon
nanostructures such as carbon nanotubes
(CNTs)[10,11] and graphene (or graphenic
flakes).[12,13] This fast growth has promoted
the development of novel materials which
offer not only improved mechanical per-
formance, but also electrical conductivity and coupled multi-
functionality, which enables the development of several kinds of
electro-mechanical sensors.[14,15]
As seen from Figure 1, the applications are vast, from
development of electromechanical devices,[16,17] to smart
textiles,[18] flexible electronics,[12] soft robotics,[19] sensors of
human motion,[20] and smart materials for in situ monitoring of
damage and structural health of advanced composite materi-
als.[21,22] Albeit the same basic piezoresistive principle prevails,
the mechanisms governing the piezoresistive behavior are quite
different and more complex for polymer composites filled with
carbon nanostructures than for continuous metallic conductors
(such as strain gages), as it will be further explained in the
following section.
2. Piezoresistivity Mechanisms and
Fundamentals of Self-Sensing of Strain and
Damage in Polymer Composites Modified with
Electrically Conductive Fillers
The piezoresistivity of a metal wire arises from the change in its
length and cross-sectional area upon mechanical loading.[1,25]
The situation is very different for insulating polymers filled with
conductive (carbon) nanostructures, which are not formed by
continuous conductors spanning the length of the material, but
comprise the summation of discrete (micron-size) elements
assembled together to form continuous paths/networks. Since
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

in these materials the matrix is non-conductive, the only way that

current can flow through the material upon voltage application Francis Avil
es is a Scientist at the
is through the formation of a network of conductive fillers Materials Department of the
(physically touching each other or not), which contain “Scientific Research Center of
delocalized electrons, see Figure 2. This phenomenon is known Yucatan” (CICY for its Spanish
as electrical percolation, and the exact filler concentration acronym), located in M
erida,
where such a network is formed is known as the percolation Mexico. He is a Mechanical
threshold.[26,27] The concentration of carbon fillers required to Engineer with a Master degree in
reach such a percolation threshold depends on many aspects Applied Physics and a PhD in solid
such as the morphology of the filler, its own conductivity, its body mechanics. He received his
dispersion state within the matrix and the rheology of the PhD in 2005 from Florida Atlantic
polymer matrix, and may range from 0.01 wt% to less than University, working on the mechanics of composite
10 wt%.[27–29] Once the percolation threshold is reached, the materials and sandwich structures, and worked as a
composite becomes electro-conductive and the change of its Research Associate at the Center for Composite Materials
effective electrical resistance upon loading can be used as a of the University of Delaware. He returned to Mexico in
piezoresistive feature to self-sense deformation and dam- 2006 and worked as scientific researcher thereafter. His
age.[30,31] It is nowaday accepted that the electro-mechanical current research interests focus on the mechanics,
(piezoresistive) sensitivity of this kind of composites is higher for electro-mechanics and multifunctional properties of
low concentration of fillers (just above electrical percolation) composite materials, piezoresistive and thermoresistive
than for composites with high filler content (i.e., composites properties of nanostructured polymer composites, carbon
with high conductivity).[29,32,33] This is because for composites nanostructures, and smart materials.
with high filler content (high number of conductive pathways,
i.e., a densely packed network), the disruption or modification of Andr
es I. Oliva-Avil
es is an early-
a single conductive path (or a few of them) has only a minor career professor at the Division of
influence on the effective electrical resistance of the composite, Engineering and Exact Sciences at
Figure 2 (top); in this figure the initial electrical resistance of the the Anahuac-Mayab University,
composite (Ra0 ) does not change much when the strain is applied located in M
erida, Mexico. He is a
(R  Ra0 ). In contrast, if the filler content is low (few conductive Physics Engineer with MS (2010)
pathways, that is, a loosely packed network with initial electrical and PhD (2013) degrees in Applied
resistance Rb0 ), the disruption or modification of a single Physics from the Center of Research
conductive path upon stretching the composite greatly affects R, and Advanced Studies (CINVESTAV
that is, upon stretching R >> Rb0 , Figure 2 (middle). Extrapolat- for its Spanish acronym). During his
ing this concept, in theory, a switching device (which switches Ph.D. he spent a year as a research scholar in the
from conductive to insulating) with initial resistance Rc0 could be Aerospace and Ocean Engineering Department at Virginia
achieved if only one conductive path would be formed within the Tech, USA. His research interests focus on the influence
insulating matrix, which conceptually represents the maximum of the structure-property relationships of carbon
piezoresistive sensitivity, as depicted in Figure 2 (bottom). nanostructured polymer composites and on the study of
However, in practice, working very close to the electrical physical properties of smart multifunctional materials.
percolation threshold generates a noisy electrical signal and
serious instrumentation drawbacks, which currently hinders the
Marco Cen-Puc is a Mechatronics
use of this very low filler concentration regime. To overcome this
Engineer by the Autonomous
practical limitation, most of the piezoresistive work on polymer
University of Yucatan (UADY) in
composites uses filler concentrations that are significantly above
M
erida, Mexico (2013). He received
the percolation threshold, where electrical resistances in the
a MS degree in Polymeric Materials
range of kΩ–MΩ are typically achieved and easily handled.
by the Scientific Research Center of
Linear piezoresistivity can be quantified by a parameter
Yucatan (CICY) in 2016. After
known as the gage (or gauge) factor, which is a metric of linear
obtaining his MS degree he worked
piezoresistive sensitivity. The gage factor (GF) is defined as the
as research assistant at the CICY’s
slope of the linear relationship between the fractional change in
Materials Department, working with
electrical resistance (ΔR/R0, where R0 is the initial electrical
Dr. Francis Avil
es on the development of temperature
resistance corresponding to the unstrained/unloaded state) and
sensors based on carbon nanotube/polymer composites.
the applied uniaxial strain. Typical metallic strain gages used in
His interests include the development and
materials testing have GFs between 2.1 and 2.3. Contrastingly,
instrumentation of sensors as well embedded electronic
the inherent piezoresistivity of carbon nanostructures can be
systems. He is currently working as a part-time professor
quite high.[34–36] For example, landmark works have reported
at the Anahuac Mayab University while he applies for a
gage factors for individual singlewall carbon nanotubes
PhD position abroad.
(SWCNTs) as high as 2900.[34] This is at least two orders of
magnitude higher than the sensitivity measured when the
nanotubes are embedded in polymer composites, which

Adv. Eng. Mater. 2018, 1701159 1701159 (2 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Figure 1. Applications of piezoresistive polymer nanocomposites based on carbon nanostructures. a) Highly twistable tactile sensing array made from
polydimethylsiloxane (PDMS) and carbon black nanoparticles.[23] (Reprinted from ref. [23], with permission from Elsevier), b) stretchable singlewall CNT/PDMS
strain sensor for human motion detection.[20] (Reprinted from ref. [20], with permission from Nature Publishing Group), c) Organic light-emitting diode on a
graphene/polymethyl methacrylate (PMMA) transparent conducting electrode.[24] (Reprinted from ref. [24], with permission from Nature Publishing Group), d)
schematic showing the use of carbon nanotubes in a polymer matrix to self-sense crack propagation and damage in advanced multiscale composites.

evidences that the full piezoresistivity of the carbon nanotube is
not transferred to its polymer nanocomposite. This is partly
because the applied strain is not fully transferred to the filler.[25]
Therefore, in this kind of nanocomposites, the inherent
piezoresistivity of the carbon fillers and the change in their
dimensions upon loading provide only a small contribution to
the measured piezoresistivity.[25,37]
In polymer composites at relatively low filler contents,
piezoresistivity is driven by the change in the filler network
configuration upon mechanical loading. The major underlying
mechanisms which govern the change in electrical resistance
upon loading are schematically represented in Figure 3. They
can be grouped into three main aspects which occur upon strain
application, viz. reconstruction of the conductive network
(creation and destruction of conductive pathways), changes in
the contact resistance between conductive fillers, and changes
in the filler-to-filler interparticle distance.[38–42] As depicted in
Figure 3, upon straining the composite new conductive pathways
can be created and others destroyed; the filler-to-filler contact
resistance also changes upon strain application, since the filler-
to-filler contact area and the number of contact points also
changes. There is also a very important non-contact scenario,
which is governed by the interparticle distance. It is accepted
that upon sufficient energetic excitation (voltage, temperature)
the electrons can tunnel or hop over small distances crossing a
thin polymeric layer; this (quantum) tunneling resistance
depends on the interparticle distance in an exponential
fashion.[43–45] The maximum tunneling distance in polymers
filled with carbon nanostructures varies depending on the
nanostructure and polymer type; the maximum tunneling
distance has been theoretical suggested to be close to
2 nm,[40,44,46] although some other studies have considered
slightly higher tunneling distances (5 nm).[45]
The relative influence of these mechanisms driving the
effective measured piezoresistivity in a polymer composite
depends on several factors related to the filler itself, to the
polymer matrix, and to their interaction and state of dispersion.
Several numerical and analytical modeling efforts have advanced
towards understanding the piezoresistive effect in polymer
composites, including effects such as tunneling, CNT-to-CNT
contact, CNT morphology and waviness, agglomeration, inher-
ent CNT piezoresistivity and loading conditions.[37,41,45,47–54]
From the materials viewpoint, the geometry of the filler and its
dispersion within the matrix, the mechanical and rheological
properties of the polymer matrix, as well as its electrical
permittivity are among the most influential factors.[54–57]

Adv. Eng. Mater. 2018, 1701159 1701159 (3 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Figure 3. Piezoresistive mechanisms in polymer composites filled with

carbon nanostructures.

distance, this capacitance may also change upon strain

application and affect the effective electrical properties of the
strained material.
Regarding the morphology of the filler, the aspect ratio
(length/diameter) is one of the fundamental factors which
govern the percolation threshold and the piezoresistive
response of nanocomposites.[27,55,59,60] The influence of this
morphological effect can be explained by using the concept
of excluded volume. [61] The excluded volume theory consid-
Figure 2. Relative influence of breaking a few individual conductive
ers that the most relevant parameter in the formation of
pathways in a network with high filler content (top), low filler content
(center) and with only one connecting pathway. Ra0 , Rb0 , and Rc0 are randomly packed percolated networks is not the actual
correspondingly the initial (unstrained) electrical resistances for cases “a” volume of the filler itself, but the volume around such a filler
(top), “b” (middle), and “c” (bottom). in which the center of another object of similar shape cannot
penetrate, that is, its excluded volume,[62] see Figure 4. The
orientational average excluded volume (V Exc ) of randomly
oriented rods,[63] infinitesimally thin disks, [59] and
Tunneling resistance is known to have an exponential depen-
spheres[62,64] are, respectively,
dence with the filler-to-filler separation distance and also with
the square root of the height of the tunneling barrier.[43] While
π   
4
the filler-to-filler separation distance is controlled by the V rExc ¼ L2r dr þ 2πd2r Lr þ πd3r ð1aÞ
excluded volume of the filler, its concentration and state of 2 3
dispersion within the polymer, the tunneling barrier is
interpreted as the thin ( < 5 nm) layer of polymer between the  2
fillers where the electrons can tunnel or hop.[43,44] Therefore, this π
V dExc ¼ d3 ð1bÞ
tunneling barrier, in turn, is strongly affected by the dielectric 8 d
properties of the polymer matrix.[43,45,53]
Other effects such as the formation of small micro-capacitors  
also strongly depend on the interparticle distance and on the 2π 3
V sExc ¼ d ð1cÞ
dielectrical properties of the thin polymer layer separating the 3 s
conductive fillers. The main idea of the micro-capacitor effect is
that each two neighboring carbon fillers separated by a very thin where the superscripts “r”, “d”,“s”, stand for rods, disks, and
polymer layer can be considered as a local microscale capacitor, spheres, respectively, Lr is the rod’s length, dr its diameter, dd is
with the fillers acting as the two electrodes and the thin polymer the disk’s diameter (lateral size) and ds is the sphere’s diameter.
layer in between as the dielectric. This effect yields a substantial For a nanotube, since Lr >> dr the first term of Equation (1a)
increase in the intensity of the local electric field around the represents a very close approximation to its excluded volume.
fillers, which promotes the charge carriers to migrate and Using this approximation (Lr >> dr) the ratio of the excluded
accumulate at the interface of the electrodes (fillers), yielding volume for rods (Eq. 1a) with respect to that of spheres (Eq. 1c)
interfacial polarization.[58] Because of the change in interparticle and disks (Eq. 1c) becomes,

Adv. Eng. Mater. 2018, 1701159 1701159 (4 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 4. Schematic representation of the concept of excluded volume (area) in a 2D
projection. a) 25 identical solid circles covering a squared area, b) 5 cylinders (whose total
area approximately equals that of the 25 circles) covering the same squared area.
 
V rExc 3 Lr 2
¼ ð2aÞ
V sExc 4 ds
! while Equation (3b) is significantly lower than 1. This means that
V rExc 4 L2r dr
¼ ð2bÞ
V dExc π d3d
The comparison between the excluded volume of CNTs and that
of spherical carbon nanoparticles such as carbon black nano-
particles or fullerenes can be done in a straightforward manner
using Equation (2a). Since the diameter of a carbon nanoparticle
(say ds  20 nm) is typically about 2 orders of magnitude smaller
than the “typical” length of a CNT (say Lr ¼ 2 μm), Equation (2a)
implies that the excluded volume of carbon nanotubes is about 4
orders of magnitude higher than that of spherical carbon
nanoparticles. A schematic of this is depicted in Figure 4 in a
bi-dimensional (2D) projection, where the total area covered by the
25 (black) solid circles (2D projection of spheres) is approximately
equal to the total area covered by the 5 elongated entities (2D
projection of cylinders); that is, Figure 4a and b have the same area
fraction of solid fillers. As seen from Figure 4a, there is still room to
fit more identical circles among the existent solid ones (these
spaces are represented as empty circles with dashed lines, to
pictorially prove the point) so the excluded area (dashed red spaces)
is very small in this case; however for the case of cylinders
(Figure 4b), assuming that interpenetration is not allowed, there is
plenty of free area that cannot be interpenetrated by another
cylindrical entity of the same dimensions, since it does not fit the
space available, that is, there is significantly more excluded area.
Notice also that, at the same area fraction, the solid circles of
Figure 4a do not form a single percolated path, while at least a
vertical contacting path can be found among the solid cylindrical
entities of Figure 4b. This is an important geometrical aspect
which explains the significantly lower percolation thresholds
achieved for elongated cylinders (carbon nanotubes) compared to
spherical particles (e.g., carbon black).[60,65,66]
Adv. Eng. Mater. 2018, 1701159 1701159 (5 of 23)
www.aem-journal.com
Comparison of excluded volumes between
CNTs (cylinders) and graphenic platelets (infi-
nitely thin disks) through Equation (2b) requires
further considerations, since graphenic platelets
can vary significantly in lateral size. For the sake
of analysis, let’s assume as a lower bound (small
sheets) for the lateral dimension of the graphenic
sheets (dd) the situation where dd is equal to the
CNT diameter (a few nanometers for SWCNTs
or tens of nanometers for multiwall carbon
nanotubes (MWCNTs), i.e., nanosheets). In this
scenario dr ¼ dd and Equation (2b) becomes,
 
V rExc 4 Lr 2
¼ ð3aÞ
V dExc π dd
On the other extreme, considering large
graphenic sheets with Lr ¼ dd (a few microns,
i.e., microsheets) yields,
 
V rExc 4 dr
¼ ð3bÞ
V dExc π dd
For CNT lengths in the order of microns and diameters in the
order of tens of nanometers Equation (3a) is much higher than 1,
the excluded volume of graphene and few layer graphene sheets
can be less or more than that of CNTs depending on the
proportion between the lateral dimension of the sheets and the
CNT length.
A high excluded volume means that the network of fillers
within the composite has many spaces among the fillers, that is,
that the network is loosely packed (see Figure 4b). This means
that the number of conductive paths is small and as such, small
changes in the network configuration (upon strain application)
are expected to produce large changes in electrical resistance, as
explained in connection with Figure 2. Thus, a high excluded
volume is expected to promote high piezoresistive sensitivity,
and this may be the reason why higher piezoresistive sensitivity
has been found for composites with CNTs as fillers, as compared
to another fillers such as spherical carbon particles or flakes of
smaller lateral dimensions at the same weight fraction.[67] It
should be kept in mind, however, that the number of filler-
to-filler contact points within the conductive network is also a
strong factor which influences the electrical conductivity and
piezoresistivity of this kind of nanocomposites. It is speculated
that a relative high number of contact points should yield
higher piezoresistive sensitivity, because of the higher changes
in contact resistance produced upon straining the composite.
Precise calculations of the number of contact points within a
network requires specialized statistical calculations, such as
those based on molecular dynamics, Montecarlo methods, or
other statistical tools.[63,68] According to Wouterse et al.[68] the
contact number for uncorrelated elongated spherocylinders
depends on several factors such as the filler concentration and
aspect ratio. According to these authors, above a critical filler
concentration, the contact number for rods increases with
increased filler volume fraction. They also found that the effect
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
of the aspect ratio is nonmonotonic, with the contact number
first increasing for aspect ratios less than six and then decreasing
for elongated spherocylinders with aspect ratios above six. For
aspects ratios above 80 the contact number is weakly
dependent on the aspect ratio but below this aspect ratio the
dependence is significant. Thus, a compromise may exist
between the density of contact points within the conductive
network, the excluded volume and the filler concentration. It is
also important to bear in mind that the perspective of excluded
volume and number of contact points covers only the
morphological and statistical packing aspects between fillers,
and many other aspects such as the rheology of the polymer and
its dielectric properties, the waviness of the filler, the filler
functional groups and filler/matrix interface (which impacts
strain transfer), alignment and dispersion, and the level of strain
attained need also to be taken into account when analyzing the
piezoresistive response of a carbon filled polymer composite.
Finally, it should be pointed out that not all changes in the
effective electrical resistance of a composite should be named
piezoresistivity. The key issue in piezoresistivity is that the
changes in electrical resistance are driven by changes in strain
(or stress/pressure) in the material. Many other aspects that
occur in advanced polymer composites such as matrix
cracking,[31] delamination,[69] or even moisture or temperature
variations[70] may also yield changes in the effective electrical
resistance of the composite, but those phenomena (albeit
interesting) should not be termed piezoresistivity. This is
because in those cases the change in electrical resistance is not
directly triggered by changes in strain or stress, but by other
physical parameter.
3. Polymer Composites Filled with Graphite
and Carbon Black
This section covers past and present work on the piezoresistive
behavior of polymer composites filled with two of the earliest
carbon allotropes investigated to this aim, viz. carbon black (CB)
and graphite. A review of the most relevant results, trends, and
findings regarding the piezoresistive response, strain, and
damage sensing of these composites is presented in this section.
The inclusion of these two fillers to
thermosettings, thermoplastics, and
elastomers are discussed as a function
of their distinctive piezoresistive re-
sponse. Furthermore, strain and pres-
sure sensing work ranges are identified,
and a summary of the piezoresistive
sensitivity (GF) for different polymer
composites is discussed. A literature
survey summarizing the piezoresistive
behavior of polymer composites filled
with CB and graphite is included in
Table A1 of Appendix A. Such a table
is organized by filler type and includes
thermosetting (TS), thermoplastic (TP)
and elastomeric (ES) matrices, as well
as the corresponding filler concentra- Figure 5. Summary of published gage factors for CB and graphite polymer composites. a) CB/polymer
tion (wt%) and loading type (tension, composites, b) graphite/polymer composites. Data extracted from Table A1.
Adv. Eng. Mater. 2018, 1701159 1701159 (6 of 23)
www.aem-journal.com
compression or bending). The main features of the piezoresistive
response of the material are captured in Table A1 by four
parameters, referring to its linear (L) or nonlinear (NL) character,
maximum strain sensing capability (emax), extend of linear
piezoresistivity (eL), and piezoresistive sensitivity (GF). The
difference between eL and emax provides an indication of nonlinear
piezoresistivity.
Carbon black (a black powder produced by the incomplete
combustion of carbonaceous or petroleum products) is likely the
earliest carbon filler that has been extensively used to produce
electro-conductive polymer composites with strain sensing
capabilities. In Figure 5, summary plots of the piezoresistive
sensitivity (GF) of CB/polymer (Figure 5a) and graphite/polymer
composites (Figure 5b) are presented (data extracted from
Table A1). An immediate observation form this table is that, in
comparison to other carbon nanostructures such as carbon
nanotubes or graphenic flakes, the concentration of CB used
for piezoresistive applications is notoriously higher. This is a
geometrical aspect, since higher filler concentrations are needed
to reach electrical percolation for spherical particles than for
nanostructures of high aspect ratio, see Figure 4. As seen from
Figure 5a, for CB, ES matrices have been more explored than TS
and TP polymers. It is also noticed that the CB concentration of
TS and TP composites is generally below 15 wt%, while higher
concentrations (>30 wt%) are used for ES matrices.
In general, the piezoresistive response of CB/polymer
composites is nonlinear, see Table A1. Gage factors are
customary extracted from the initial part of the piezoresistive
curve (i.e., e < 5%), where an approximately linear behavior is
observed. The ultimate strain of composites with TS matrices is
usually low ( < 6%), while TP and ES composites can sustain
larger strains. The majority of the piezoresistive works report
uniaxial tension loading, see Table A1, and the majority of the GF
reported range between 3 and 15, see Figure 5a. Negative
piezoresistivity has also been observed under tension and
compression.[38] especially for elastomeric composites. Negative
piezoresistivity means that, under uniaxial tensile loading (e>0),
the electrical resistance decreases (ΔR < 0), or that ΔR increases
under compression. For TS composites, the family of CB/epoxy
under tensile loading is the most investigated one, with ultimate
strains below 6% because of the brittleness of epoxy in
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
tension.[71,72] The piezoresistivity of CB/TP composites has also
been investigated,[73–78] finding positive GFs between 3 and 30,
see Figure 5a. Zhao et al.[74] evaluated the tensile piezoresistive
behavior of a 13 wt% CB/polypropylene composite and
reported a nonlinear increasing piezoresistive behavior. The
authors argument that both, formation and breakdown of
conductive paths, compete during the loading process, making
the curve nonlinear. As seen from Table A1, CB/ES composites
made from rubber or polyisoprene are among the most common
composites which have been investigated using CB.[73,79–85]
Styrene ethylene butylene styrene (SEBS), a thermoplastic
elastomer, has also been filled with CB presenting large
deformation capabilities and piezoresistive strain sensing.[86,87]
Elastomers are by far the most explored polymer matrices for CB
likely because of their large stretch ratio, high flexibility, and
interest of the rubber industry. However, they are also more
prone to present time-dependent viscoelastic relaxation and
creep, which is captured by the piezoresistive response.[87,88] For
elastomeric composites under tension, most of the reviewed
literature (Table A1) shows an increase of electrical resistance
with increased strain, that is, positive piezoresistivity, following a
nonlinear behavior, especially at high strains (above 20%). When
the composite is stretched, CB agglomerates are pulled away,
disrupting certain conducting paths, with the concomitant
increase in R. According to Figure 5, the GFs (extracted from
e < eL, which ranges from 4% to 17% depending on the material)
for CB/ES composites have been reported above 12. A different
piezoresistive behavior under tension has been observed by
other authors such as Flandin et al.[8,38] for an ethylene-octene
elastomer filled with CB. These authors reported that, under
tensile loading, after the increase in R during the first stages of
strain (elastic region) a decrease in R (negative piezoresistivity)
occurs at higher tensile strains. The authors claim that for the
ethylene-octene (comprising flexible chains with fringed micel-
lar crystals), the large strains produce new conducting pathways
or improves the existent ones by CB reorientation and/or strain-
driven agglomerate breakdown. Both positive and negative GFs
have also been reported in different works under compression,
see Figure 5a. Negative GFs, that is, increase of R with increased
compressive strain, have been explained in terms of the
destruction of the conductive network under compression,[84,85]
while positive GFs have been explained in terms of increased
particle volume fraction in the composites due to volume
reduction during compression.[73] Wang et al.[89] explored CB/
silicone rubber composites at CB-to-rubber mass ratios varying
from 0.08:1 to 0.24:1 (7.4 to 19.4 wt%). These authors report both
positive and negative piezoresistivity for the same specimen
at different levels of strain, and also depending on the CB
concentration, see Figure 6. For example, for CB concentrations
above 11.5 wt% they found positive piezoresistivity for all strain/
stress levels tested, but between 9.1 wt% and 10.7 wt% they
found both, positive and negative piezoresistivity, see Figure 6.
For concentrations of 8.3 wt% and below, however, the
piezoresistive behavior turned monotonically negative, see
Figure 6. This complex behavior was explained in terms of
three main effects caused by the translation and rotation of the
CB particles due to compression, viz. reduction of the gap
between conductive paths, formation of new conducting paths,
and destruction of existing conductive paths. The first two
Adv. Eng. Mater. 2018, 1701159 1701159 (7 of 23)
www.aem-journal.com
mechanisms support positive piezoresistivity, while the latter
one explains negative piezoresistivity.
Several authors have also explored the piezoresistive behavior
of CB/polymer composites under cycling loading,[8,38,72,74–
77,85,90]
viscoelastic relaxation,[87] and creep,[73,84] and a few of
them have pointed out the effect of hysteresis.[72,74,75,77,90]
Damage and irreversible phenomena can also be assessed by
means of the electrical response of the composite. In this regard,
CB/HDPE composite films under uniaxial tension have shown
negligible variations in R for e < 8% (elastic region, according
to the authors), and a sudden increase of R for higher strains
corresponding to the elastic-to-plastic transition.[76] The authors
state that this unique electrical property could be used to identify
the elastic and plastic regions, representing a valuable tool for
damage assessing.
Regarding the use of graphite as a filler material, the
piezoresistive behavior of graphite/polymer composites
under tension,[91] compression,[6,92–94] and bending[95,96] has
also been investigated. Figure 5b summarizes the GFs of
graphite/polymer composites for different polymer matrices.
In comparison to CB/polymer composites, evaluation of the
piezoresistivity of graphite/polymer composites in terms of
strain or GFs has been less explored. A great number of the
existent works report on sensitivity as a function of applied
pressure, that is, calibrating as a pressure sensor. According to
our literature survey (see Table A1 and Figure 5b), the great
majority of graphite/polymer composites exhibit positive
piezoresistivity with the majority of GFs ranging from 7 to
25. From the literature survey conducted, Shevchenko et al.
reported the highest GFs for graphite/polymer composites using
a polypropylene (PP) matrix under tension, with GFs associated
to the linear region of the piezoresistive plots (e < 1%) ranging
from 7 (45 wt%) to 100 (12.5 wt%).[91] These authors also
presented evidence that the linearity of the graphite/PP
piezoresistive curves increases as the graphite volume fraction
increases. Serra et al.[96] reported on the piezoresistive behavior
of graphite/epoxy composites under bending. The composites
were bonded to a rigid substrate (alumina or aluminum) which
Figure 6. Piezoresistive response of CB/rubber composites for different
CB loadings.[89] (Reproduced from ref.[89], with permission from
Elsevier).
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Table 1. Summary of pressure-dependent piezoresistive behavior of polymer composites filled with CB and graphite.

Pressure-dependent piezoresistivity
Filler wt% Matrix L/NL Pmax (kPa) PL (kPa) S (x103, kPa1) Ref.
[99]
Carbon black 20 Natural rubber NL 2400 1600 0.4
[100]
10.3 PMVS L 1200 1200 0.07
10 Silicon rubber NL 600 80 1.35 to 1.65 [101]

8 Silicon rubber L 700 700 –0.4 [102]

5 Silicon rubber NL 1000 800 2.5 [103]

0.8 Silicon rubber NL 1000 500 0.25 [89]

[89]
0.16 Silicon rubber NL 1000 500 0.5
10 Polyisoprene NL 400 40 0.5 [104]

10 Polyisoprene NL 100 40 3 [105]

[106]
Graphite 0.5 Epoxy NL 75 000 13 000 0.05
20 HDPE NL 25 000 5000 0.04 to 0.08 [107]

[106]
3 Silicon rubber NL 750 200 1.3
6.6 Silicon rubber NL 500 100 0.3 [108]

[108]
22.1 Silicon rubber NL 500 100 1.2
[98]
30 Polyurethane NL 60 5 63

Acronyms: HDPE ¼ High density polyethylene, PMVS ¼ poly(methylvinylsilioxane), L ¼ Linear, NL ¼ Nonlinear, Pmax ¼ Maximum pressure registered by the electrical
signal, PL ¼ Maximum pressure corresponding to approximately linear piezoresistivity, S ¼ Pressure-dependent piezoresistive sensitivity.

was directly bent. Authors found that the GF of graphite/epoxy as the number of cycles increases. These authors also
composites bonded to a beam decreases with increased graphite highlighted the dependence of the piezoresistive response on
concentration for both alumina and aluminum substrates. the creep behavior of the polymeric matrix and the elastic
Using a similar bending approach, Frazier and Allen[95] reported properties of the fillers.[6] Other authors such as Chen et al.[94]
the piezoresistive behavior of graphite/polyimide composites have also reported that the hysteresis reduces with increased
bonded to a rigid silicon substrate. These authors found that the number of cycles for graphite nanosheets in silicone rubber.
GF of the composites first increases with increased graphite For pressure sensors, the piezoresistive behavior of CB-based
concentration, and then decreases, reporting a peak GF of 16.8. It and graphite-based polymer composites can also be approached
is important to notice that when the piezoresistive material is in terms of applied pressure (instead of applied strain). To
not directly stretched but bonded to a beam, the stress/strain quantify the linear piezoresistive sensitivity as a function of
transfer from the tested beam (substrate) may not be perfect,
which would affect the piezoresistive measurement and
determination of gage factor. On the other hand, Beruto
et al.[93] reported that R decreases with increased compressive
strain (i.e., positive piezoresistivity) for graphite/rubber compo-
sites, being such changes more significant as the graphite
concentration decreases and approaches the percolation thresh-
old. Chen et al.[94] reported a negative piezoresistivity for a
graphite nanosheet/rubber composite under compression,
which is also more pronounced as the graphite concentration
is closer to the percolation threshold. The authors argument that
the mechanism underlying such a negative piezoresistivity is
the breakdown of conductive pathways and that the transport
mechanism can be understood in terms of the tunneling
conduction between conducting particles. Cyclic load-
ing[6,91,94,96] and relaxation tests[92,96] have also been explored
by a few authors. Regarding irreversible effects, Shevchenko
et al.[91] showed that the piezoresistive hysteresis under loading-
unloading cycles is reduced with increased number of cycles and
by increasing the particle concentration. Lu et al.[6] also reported
that the piezoresistive response is different among sequential Figure 7. Summary of published gage factors for SWCNT and MWCNT
loading-unloading cycles, being more stable (but less sensitive) polymer composites. Data extracted from Table A1.

Adv. Eng. Mater. 2018, 1701159 1701159 (8 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
the applied pressure (P), a pressure-dependent piezoresistive
coefficient S ¼ (ΔR/R0)/ΔP has been customarily used.[97,98]
This coefficient is the pressure analogous of the GF, which is
defined in terms of pressure instead of normalized strain.
Table 1 summarizes the most relevant pressure-dependent
piezoresistive parameters for several CB and graphite polymer
composites. Relatively high filler loadings, usually above 5 wt%,
are generally used for these applications, although composites
with lower concentrations (<1 wt%) can also be found, see
Table 1.
Higher loadings of CB and graphite, as compared to other
fillers such as carbon nanotubes or graphenic flakes, are
required for pressure sensing applications of these materials due
to the corresponding lower excluded volume and aspect ratio of
these fillers. Elastomeric matrices are undoubtedly the most
used ones for pressure sensors due to their suitable flexibility
and mechanical properties. The applied pressures customarily
range from few tents of KPa to nearly 100 MPa. Except for the
high value (63
103 kPa1) reported by He et al.,[98] the
magnitude of the reported sensitivity factors in Table 1 ranges
from 0.04
103 to 3
103 kPa1. Positive and negative
pressure-dependent piezoresistivity has also been observed.
4. Polymer Composites Filled with Carbon
Nanotubes
Among the carbon-filled polymer composites, those modified
with CNTs are undoubtedly the ones that have captured more
attention of the scientific community, as evidenced by the
numerous publications to date. They range from the early work
on buckypaper of Dharap et al. in 2004[109] to current and
ongoing research.[110,111] A brief literature survey summarizing
the piezoresistive behavior of polymer composites filled with
SWCNTs and MWCNTs is included in Table A1 of Appendix A.
Such a table is organized by filler and matrix types and includes
the corresponding filler concentration (wt%) and loading type, as
discussed in section 3.
The literature survey suggests that the majority of published
works are on epoxy composites filled with MWCNTs, and
substantial work has also been published using a variety of
thermoplastic and elastomeric matrices such
as thermoplastic-polyurethanes (TPUs) and
polydimethylsiloxane (PDMS). By far, the
great majority of the piezoresistive works
report only tensile loading, being compres-
sion and flexural loading significantly less
reported.
Figure 7 summarizes the piezoresistive
sensitivity (in the form of gage factors) of the
SWCNT and MWCNT polymer composites
listed in Table A1. Data for buckypapers (BP) is
also included for completeness. As seen from
this figure, the great majority of GFs for CNT
polymer composites are found between 1 and
7, without a particularly distinctive trend Figure 8. Piezoresistive response of thermosetting polymer composites. a) Linear response of
between SWCNTs and MWCNTs. However, a 0.3 wt% MWCNT/vinyl ester composite.[121] (Reproduced from ref. [121], with permission from
composites with MWCNTs seem to be more Elsevier), b) nonlinear and nonmonotonic response of a 0.1 wt% MWCNT/epoxy composite.[71]
reported, probably because they are easier and (Reproduced from ref. [71], with permission from the American Physical Society).
Adv. Eng. Mater. 2018, 1701159 1701159 (9 of 23)
www.aem-journal.com
more economical to synthesize in large scale. Notice that the
magnitude of the GFs for SWCNT and MWCNT composites in
Figure 7 are comparable to those obtained using graphite or
carbon black in Figure 5, but as seen from Table A1, those for
graphite and carbon black require significantly higher filler
concentration. Reports on negative gage factors are scarce, but
they have also been reported, especially under compressive
loading.[112] Particular features of CNT polymer composites with
TS, TP, and ES matrices are discussed in the following sections.
4.1. Thermosetting Matrices
Epoxy thermosettings filled with SWCNTs[113] and
MWCNTs[48,71,72,110,114–117] are among the most extensively
investigated polymers regarding piezoresistivity of polymer
composites, compared to other thermosetting polymers such as
vinyl ester[112] or polyester. Given their low processing viscosity, a
distinctive feature of these polymer composites is their capability
of reaching electrical percolation at CNT concentration levels
significantly smaller than those for thermoplastics or elasto-
mers, see Table A1. Because of these, CNT concentrations of
0.3 wt% or below are sufficient to guarantee piezoresistivity in
thermosetting composites (see Table A1). This very low filler
concentration, alongside with the superior mechanical proper-
ties of thermosetting composites, makes them particularly
attractive for further developments of advanced multiscale
hierarchical composites loaded in tension, compression, or
bending.[31,118–120]
Linear piezoresistivity is quite common for CNT thermoset-
ting composites loaded in tension with low ultimate strains
(Figure 8a), but nonlinear and even nonmonotonic piezoresis-
tivity may also occur for thermosettings or modified thermoset-
tings with relatively “high” (within the context of
thermosettings) ultimate strains.[71] For this later case, an
inversion of the piezoresistive trend (from positive to negative)
may occur for certain CNT weight concentrations under
relatively large deformations (Figure 8b, see ref. [71]), presum-
ably because of polymer inelastic behavior. Negative piezor-
esistivity means that the effective (three-dimensional) distance
between CNTs is reduced as the polymer is axially stretched,
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
and may be caused by matrix plasticity, lateral CNT motion by
Poisson’s effect and/or three-dimensional CNT network rear-
rangement under large strains.[71,111]
The CNT weight concentration also affects the piezoresistive
response, and is commonly accepted that lower CNT concen-
trations yields higher piezoresistive sensitivities.[112,114] This is
because lower CNT concentrations yields less redundant paths
forming the CNT network, and also the tunneling effect is
expected to be more significant for CNT concentrations close to
percolation.[53]
Tensile loading is by far the most commonly investigated
loading case, although a few works have also investigated
compression[112,121] and bending.[110] Ku-Herrera and Avil
es[121]
used 25.4 mm long prisms of 12.5 mm square cross section to
investigate the piezoresistive response and damage accumula-
tion of 0.3 wt% MWCNT/vinyl ester specimens under uniaxial
compression loading. They found a nonlinear and nonmono-
tonic behavior under monotonic compression loading up to
failure (see Figure 9a), with the electrical resistance first
decreasing (positive gage factor) and then increasing (negative
gage factor) for relative large strains (e>2.5%). The transition
zone (from positive to negative piezoresistivity) was shown to
correlate with polymer yielding, pinpointing that the MWCNT
network is not only able to self-sense strain, but also polymer
yielding and plasticity.
Under cycling compressive loading at incremental strains of
magnitude ecyc (Figure 9b), a reversible behavior of ΔR/R0 was
observed for strains below the polymer yield strain, but ΔR/R0
was irreversible and cumulative for cycling strains above
polymer yielding. As seen from Figure 9b, the irreversibility
of ΔR/R0 upon unloading is an indication of plasticity and
damage accumulation, which can be quantified based on such a
permanent electrical response.[121]
Regarding flexural loading, the scarce studies seem to indi-
cate that the piezoresistive response is more sensitive under
uniaxial tension than under flexural loading,[110] although more
research is needed to unveil the reasons for that. The use of
piezoimpedance (instead of piezoresistance) is also an interest-
ing avenue that has been suggested by very few authors. It is
argued that at high frequencies (1 MHz) alternating-current
measurements (piezoimpedance) are more sensitive than those
in direct current (piezoresistance) due to the combined action
Figure 9. Electrical response of a 0.3 wt% MWCNT/vinyl ester composite under axial compressive loading. a) Monotonic loading up to failure, b) cycling
loading, each cycle reaching a strain ecyc.[121] (Reproduced from ref. [121], with permission from Elsevier).
Adv. Eng. Mater. 2018, 1701159 1701159 (10 of 23)
www.aem-journal.com
of the effective resistance and capacitance that determine the
overall electrical response of the material.[110]
4.2. Thermoplastic Matrices
Thermoplastics are generally tougher and more deformable than
thermosetting polymers; therefore, piezoresistivity has been
extended to larger strain sensing capabilities for CNTcomposites
using thermoplastic matrices (see emax listed in Table A1 of
appendix A). As seen from Table A1, a large variety of
thermoplastic polymers such as polymethyl methacrylate
(PMMA),[32,122] polyvinylidene fluoride (PVDF),[123–127] polycar-
bonate (PC),[128] polyethylene oxide (PEO),[129] polysulfone
(PSF)[130] polyethylene terephthalate (PET),[131] and polypropyl-
ene (PP)[33,74] have been modified with CNTs to achieve
piezoresistive strain sensing capabilities. In general, the gage
factors reported for CNT/thermoplastic composites in the elastic
regime are similar (or slightly above) than those for thermoset-
tings, as pointed out in Figure 7. However, because of their larger
deformation capabilities, higher gage factors than those for
CNT/thermosetting composites have been found for CNT/
thermoplastic composites in the plastic regime.[55] This larger
sensitivity is concomitant with a nonlinear piezoresistive curve,
which is usually driven by inelastic deformations. For example,
Zetina-Hernández et al.[33] found that for 4 wt% MWCNT/PP
composites, the tensile piezoresistive curve attains a gage factor
of 3.6 for strains below 1%, and the sensitivity is highly
increased (GF  28.9) for larger strains, Figure 10. An indication
of the increased sensitivity with respect to thermosetting
polymers is seen in the maximum values of ΔR/R0 attained
in Figure 10a (280%), compared to those in Figure 8a (3.3%).
The authors further conducted loading-unloading tensile
experiments to prove that the transition zone (1% < e < 2%)
corresponds to polymer yielding, and that the high gage factors
observed for e > 2% correspond to polymer plasticity. Nanni
et al.[131] also confirmed that the yield point of the polymer (PET
in this case) can be identified by a transition zone with reduced
sensitivity in the electro-mechanical signal, as seen in
Figure 10b. Using 1, 2, and 4 wt% MWCNT/PET composites,
they demonstrated a strong correlation between polymer
yielding and the electrical resistance, whose particular features
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 10. Piezoresistive behavior of CNT/thermoplastic composites. a) MWCNT/PP at 4 wt%.
(Reproduced from ref. [33], with permission from Springer), b) MWCNT/PET at 2 wt%.[131] (Reproduced
from ref. [131]
, with permission from Elsevier).
depend on the MWCNT concentration. At PET yielding
(e  4.2%), the excessive deformation of the polymer causes
significant rearrangement of the CNT network, pinpointing a
local change in trend of the electrical resistance; right after
yielding the CNT-to-CNT separation largely increases triggering
high values of ΔR/R0, see Figure 10b.
Given the large deformation and nonlinear behavior of CNT/
thermoplastic composites, electron tunneling has been frequently
cited among the major mechanisms driving piezoresistivity in
these materials.[55,132] This explanation seems particularly plausi-
ble for large deformations, where the CNT-to-CNTseparation is in
the order of a few nanometers and the form of the piezoresistive
curve is exponential, as described by the tunneling model.[43]
The behavior of the electrical resistance under cycling (tensile)
loading has also been investigated for CNT/thermoplastic
composites.[32,74,128,133] The results suggest that these smart
materials are able to self-sense strain, instantaneous, and
accumulated damage. Under cycling experiments, the ΔR/R0
curve accurately follows the applied strain, but only if such
strains are elastic. Given the inherent polymer viscoelasticity,
the piezoresistive response may become time-dependent.[127] If
plasticity, any form of damage, or viscoelastic phenomena occur,
the electrical resistance does not exactly follow the cyclic applied
strain (see Figure 11) and the extent of such a difference
correlates with the amount of inelastic strain and/or accumu-
lated damage. As for thermosettings, the capability of self-
sensing strain and damage of these composites can be exploited
for the development of advanced multiscale hierarchical
composites comprising an engineering fiber, carbon nanotubes,
and a thermoplastic matrix.[21]
4.3. Elastomeric Matrices
Piezoresistivity in elastomeric polymer composites is quite
unique because of the extremely large strain sensing capabilities
rendered, see emax for different polymers in Table A1. Because of
such large strain sensing capabilities, the piezoresistive response
of elastomeric composites is customary nonlinear, see for
example, Figure 12. [134]
Unlike thermosetting and thermoplastic nanocomposites,
the intended piezoresistive applications of elastomeric polymer
Adv. Eng. Mater. 2018, 1701159 1701159 (11 of 23)
www.aem-journal.com
composites are more leaned towards
low load bearing capacity applications,
such as stretchable and wearable
strain sensors, sensors of human
motion, soft skins, biomedical appli-
cations, and flexible electronics,
among others.[15] The common fea-
tures in all these applications are the
need of very large strain sensing
(typically e>1), while less importance
is devoted to their mechanical load
[33] bearing capacity. This is contrasting
with the aimed applications of ther-
mosetting and some thermoplastic
carbon-based composites for struc-
tural health monitoring of advanced
composites,[21,31,118,119] where the me-
chanical loads are of major concern. Thermoplastic polyur-
ethanes,[111,134–138] polydimethylsiloxanes (PDMSs),[20,139–141]
silicone rubber,[142] SEBS,[143,144] styrene butadiene styrene
(SBS),[145–148] ethylene propylene diene, [149] and natural
rubber[67] are among the major elastomers (or thermoplastic
elastomers) that have been investigated regarding piezoresistive
and strain sensing applications, see Table A1. The typical strain
sensing capabilities of these polymers range from e ¼ 50% to a
few times its initial length (1x to 3x), although strain sensing
capabilities up to 10x (1000%) have been reported,[136] as seen
from Table A1. Tunneling is invariable listed as one of the major
mechanisms contributing to the piezoresistivity of elastomeric
polymer composites,[67,132] especially at large strains. The high
nonlinearity of the piezoresistive signal for elastomeric
composites is indicated in Table A1 by the null value of eL or
by the large difference between emax and eL. This nonlinearity
yields a dependence of ΔR/R0 with strain, that is, “gage factors”
that are not constant but depend on the strain level.[134,145,146]
Therefore, very large piezoresistive sensitivities can be obtained
for elastomeric composites in the high deformation regime, see
Figure 11. Behavior of the electrical resistance under 10 loading-
unloading cycles for 3.4 vol% MWCNT/PP composites.[74] (Reproduced
from ref. [74], with permission from Elsevier).
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 12. Piezoresistive behavior of a 6 wt% MWCNT/TPU nano-
composites.[134] (Reproduced from ref. [134], with permission from John
Wiley & Sons).
for example Ref.[111,146] as indicated in Figure 7 and Table A1.
For very large strains (typically e>1), the inter-particle distance
becomes so high that the tunneling effect is no longer active,
yielding a disconnection of the conductive network and a sharp
decrease of electrical conductivity, losing the strain sensing
capability.[137] It has been shown that the strain at which this
electrical transition happens (indicated by a vertical arrow
around e  107% in Figure 12) depends on several factors such
as the polymer and CNT types, the state of dispersion and the
[134,137]
CNT concentration.
TPU composites are a special case of thermoplastic polyure-
thane elastomeric composites that have been used for large
strain sensing applications. The microstructure of CNT/TPU
composites comprises a phase separation (domains) between
the rigid and flexible segments of the polymer, with the
CNTs preferentially bonded to the urethane groups of the rigid
segment (RS) domain.[111,150] Both, the CNT content and the RS
content promote a rigid-to-flexible phase separation, forming
domains. Using MWCNT/TPU composites of 15, 30, and 50 wt%
RS content at MWCNT contents of 2, 4, and 6 wt%, Lozano-Perez
et al.[111] showed that the RS content is more influential than
the MWCNT content for achieving phase separation. Since
MWCNTs are preferentially dispersed within the RS phase
domain, it has been shown that the piezor-
esistive response of MWCNT/TPU compo-
sites strongly depends on the relative
concentration of rigid-to-flexible segments
and also on the MWCNT content.[111] As seen
from Figure 13, changes in electrical resis-
tance above 10,000 times the unstrained
electrical resistance at e ¼ 100% were achieved
for very flexible composites comprising 15 wt
% RSs and 4 wt% MWCNTs, while such
changes were about 100 times for composites
with 30 wt% RSs at the same MWCNT
concentration.
Elastomers are more prone to present
viscoelastic behavior (relaxation, creep) and Figure 13. Piezoresistive behavior of MWCNT/TPU composites at different rigid segment and
hysteresis under cyclic loading, and these MWCNT contents. a) 15 wt% RS, b) 30 wt% RS.[111] (Reproduced from ref. [111], with permission
time-dependent phenomena are reflected in from John Wiley & Sons).
Adv. Eng. Mater. 2018, 1701159 1701159 (12 of 23)
www.aem-journal.com
their piezoresistive response.[87,146,151–153] In fact, it has been
observed that the piezoresistive sensitivity of thermoplastic
elastomer composites such as SBS may depend on their strain
rate.[146] An important feature of elastomeric composites is that
they tend to present significant piezoresistive hysteresis under
cycling loading,[15,138,154] which is very relevant for any practical
implementation. This means that the pathway followed by the
ΔR/R0 versus e curve for loading is different to that for
unloading, which has been quantified by the normalized area
under the loading-unloading ΔR/R0 versus e curve, and termed
piezoresistive hysteresis (HP), see Figure 14. According to
Lozano-P
erez et al.,[153] for MWCNT/TPU composites HP also
depends strongly on the rigid segment (and MWCNT) content,
as seen from Figure 14a and b for 4 wt% MWCNT composites.
Notice also from these figures that after unloading, the electrical
resistance does not return to the same initial (unloaded) value,
which is a form of point hysteresis (as opposed to that of
trajectory), of high relevance for any practical implementation.
Lozano-P
erez et al.[153] also noticed that HP decreases as the
material is repeatedly stretched an unstretched, that is, as
the number of cycles increases, Figure 14c. This is because the
flexible and rigid domains of the segmented polyurethanes
rearrange during the cyclic deformation process, with the
MWCNT network acting as physical cross-linkers.
5. Polymer Composites Filled with Graphenic
Sheets
The interest on graphene and its few-layer derivatives has
tremendously grown since the landmark work of Geim and
Novoselov’s group reporting on mechanical isolation of an
individual graphene sheet in 2004.[155] Since then, many
research efforts have taken advantage of the electrical and
electronic properties of this family of graphenic materials
(monolayer, few-layer, and multilayer) for the development of
electro-mechanical, strain sensing, and other sensory and/or
actuating devices.[156–158] Graphene (a monolayer) has excellent
physico-chemical properties, such as a very high elastic modulus
(200 times stronger than steel), excellent electrical and thermal
conductivity, and very large specific surface area.[155,159] The term
graphene, strictly speaking, refers to a single layer of sp2 bonded
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 14. Piezoresistive response and hysteresis under tensile cycling loading for MWCNT/TPU
composites at 4 wt% MWCNTs and different RS content. a) First cycle, b) fifth cycle, c) piezoresistive also yielded very high GFs such as 150[169]
hysteresis (HP) as a function of the cycle number.[153] (Reproduced from ref. [153], with permission
from Elsevier).
carbon atoms,[160] which can be produced by bottom-up or top-
down approaches.[161,162] A host of literature has inaccurately
used the term “graphene” to refer to any member of the
graphenic family in a generic way (see Bianco et al.[160]),
downplaying the role of the number of layers and nanostructure
in the resulting effective properties of the nanocomposite. Given
this issue, international efforts to normalize the terminology of
graphene-based materials have suggested the use of a
standardized nomenclature which considers the number of
graphene layers, average lateral dimensions, and the carbon-
to-oxygen ratio as the main parameters for naming such
nanostructures.[160,163] This proposed nomenclature will be
followed in this review, supporting the international efforts to
converge to a standardized nomenclature. Given the difficulty
of maintaining thermodynamically stable individual sheets of
sp2 bonded carbon atoms inside a bulk polymer, the two-
dimensional forms of carbon filler that have been mostly used in
polymer composites are actually few-layer graphene (FLG, 2–5
layers), multi-layer graphene (MLG, 6–10 layers) or exfoliated
graphite (EG). Their distinction from work to work can be quite
difficult for a review paper, since not only they are frequently
not properly named but also their morphological (number of
layers, lateral dimensions) and physico-chemical properties are
in many instances not provided. This issue prohibits a proper
examination of structure-property relationships in this kind of
nanocomposites to date.
Free standing (single-layer) graphene has been reported as a
material with piezoresistive properties, suitable for the develop-
ment of electro-mechanical devices and strain sensing applica-
tions.[164,165] The GF of suspended single-layer graphene has
been reported as high as 151[35] and GFs of FLG films with
Adv. Eng. Mater. 2018, 1701159 1701159 (13 of 23)
www.aem-journal.com
different thicknesses and conductivities
have been reported varying from 10 to
300.[166] However, other authors have also
reported much lower GFs for FLG sheets
aimed for strain/pressure sensing, viz. 10
to 15 in ref. [167], 1.9 in ref. [164], and 2.9
to 4.3 in ref. [165]. The high GFs reported
for individual graphene or FLG sheets
greatly reduce when such a family of
graphenic sheets (GSs) are deposited on
substrates or used as fillers for polymer
composites. Pressure and strain sensing
devices based on MLGs deposited on
silicon membranes have shown GFs of
1.6,[168] while FLG patterns deposited on
polymer substrates have reported GFs
around 6.[158]
A common sensory application given
to this family of GSs consists in deposit-
ing them over a flexible substrate and
bonding the film material (or transfer-
ring the GS) to a beam for piezoresistive
strain measurement. This procedure has
and 250.[170] Direct deposition of EG on a
PMMA beam has also reported a GF of
50.[171] Other works about piezoresistive
sensors based on GS in the form of
isolated membranes, sheets or “graphene-paper” (GP) bonded to
polymeric substrates have shown GFs as high as 546 for the case
of a PDMS substrate/beam, as indicated in Table A1. This
information is summarized in Figure 15 (labeled as GP,
corresponding to 100% weight content), along with the GF of
polymer composites filled with a variety of GSs.
Regarding polymer composites filled with GSs, the filled
composites are either directly tested or bonded to a beam, and
such a beam is loaded. Direct testing of GS/polymers has been
conducted under tension[61,172–174] and compression load-
ing.[175–177] Bonding the GS/polymer composite material to a
metallic, polymeric, or composite beam for subsequent testing of
the beam has also been reported under tension[178] and, more
frequently, bending loading.[179–182]
As seen from Table A1 and Figure 15, a variety of polymer
matrices have been used for the development of piezoresistive
nanocomposites based on GSs, from thermosettings such as
epoxy[172,179,183] and vinyl ester,[181] to elastomers such as
PDMS,[174,178,182] TPUs,[176,177] and ethylene-butene copolymer
(EB).[61] According to the literature review, uniaxial tension has
been the most investigated loading scenario for this kind of
polymer nanocomposites, followed by bending tests and, less
commonly, direct uniaxial compression. As for the case of CNTs,
the reported works on polymer nanocomposites filled with
graphenic sheets also conclude that the sensitivity of the
nanocomposites is influenced by the filler concentration, with
higher sensitivity for lower filler content.[178,180,182]
Regarding GS nanocomposites with thermosetting matrices,
nonlinear behavior of the piezoresistive curve is observed for the
majority of the reviewed publications, which applies for both,
tensile and bending tests. This is a bit different to the case of
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 15. Summary of published gage factors for polymer composites
filled with graphenic sheets. Data extracted from Table A1.
CNT filled composites, where a large linear region is frequently
observed for thermosetting polymers. In general, the maximum
strain attained for all revised works with thermosetting matrices
is emax  3%.
The reports available in the literature for thermoplastic
matrices filled with the family of GSs are substantially less.
Among the few existent reports, a strain sensor using reduced
graphite oxide into a polyvinylidene fluoride (PVDF) matrix
bonded to an aluminum beam and loaded in tension showed a
GF of 12.1.[184] PVDF composites using reduced FLG sheets have
also been produced and bonded to a flexural beam, showing
gage factors between 7 to 11 when the composite is bent with
the beam.[185] This work also highlighted the influence of the
morphology of the carbon filler, yielding different gage factors
for FLG oxide, reduced FLG sheets, and carbon nanohorns.[185]
Three-dimensional reduced few layer graphene oxide has also
showed excellent flexibility and piezoresistive properties when
mixed with a polyimide matrix to fabricate sponge materials.[186]
Polystyrene has also been used as a matrix for composites filled with
GS tested under compression showing self-sensing of strain.[187]
Similar to the case of CNTs, GS composites show higher GFs
for elastomeric matrices than for any other polymer matrix (see
Figure 15). This is expected due the flexibility and large straining
capabilities of elastomeric polymers, which render high GS
mobility. However the piezoresistive behavior of such nano-
composites typically exhibit a highly nonlinear (and time-
dependent) behavior[61,174–177] along with high emax, see
Table A1. Similar to CNT composites, their high deformation
capabilities makes them suitable for development of strain
sensors for human motion, electrodes for flexible electronics,
wearable electronics and human/computer interfaces.[188–190]
GS/polymer nanocomposites not only have the ability to self-
sense strain, but also damage and other irreversible phenomena.
This has been more explored for thermosetting polymers, given
their higher stiffness and strength, which makes them more
suitable for load bearing applications. Zha et al.[172] reported
the application of 2.7 wt% amino functionalized GS/epoxy
Adv. Eng. Mater. 2018, 1701159 1701159 (14 of 23)
www.aem-journal.com
composites for structural health monitoring under tensile
loading. As indicated in Figure 16, the electro-mechanical curve
of such nanocomposites showed two zones; the first one
presented a quite linear increment of fractional electrical
resistance, while the second one showed abrupt incremental
steps of electrical resistance. The first (linear) region can be
correlated to the elastic deformation of the material, while the
incremental steps in the second zone are attributed to
accumulation of microcracks and permanent damage in the
nanocomposite.[172] Proper functionalization of carbon nano-
structures is known to improve their dispersion and load transfer
with the polymer matrix, but its direct effect on the piezoresistive
sensitivity is yet debatable.
Similarly, following ideas previously proposed for CNTs, a few
other works have extended this concept using graphenic sheets
for in-situ structural health monitoring of fiber reinforced
polymer composites, by adhering the graphenic nanocomposite
to tensile or bending beams.[191] Cyclic tests of graphenic
composites have also indicated residual changes in electrical
resistance after cycling loading,[172,180,191,192] caused by rear-
rangement of the conductive pathways during cyclic excursions.
An interesting feature observed from comparison of the GFs
in Table A1 for graphenic composites to those for CNT
composites, is that most GFs for composites using CNTs as
fillers are below 30 (see Figure 7), whereas those for composites
using GSs are overall higher (the span is almost twice, according
to the inset in Figure 15). This trend may be explained in terms of
the excluded volume, and indicates that the aspect ratio and
morphology of conductive fillers play a paramount role in the
piezoresistive response of nanocomposites,[61,67] as explained in
section 2. These data indicate that strain sensitive composite
materials with higher sensitivity could possibly be developed by
the use of two-dimensional graphenic sheets. As seen from
section 2, this of course depends on the relative dimensions of
the fillers (see Eq. 3), and other factors such as the strain level.
However, larger piezoresistive sensitivity has also been achieved
for composites with CNTs than with GSs.[67] Another feature
observed from Table A1 is that the filler concentration used for
such piezoresistive applications is significantly higher for GS
than for CNT composites. Besides, GS/polymer composites also
exhibit noticeable nonlinearity and strain sensing capabilities
Figure 16. Electro-mechanical response of 2.7 wt% EG/epoxy compo-
sites.[172] Reproduced from ref. [172] with permission from Elsevier.
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
which depend on the deformation level.[67,180,191,192] This
features may not be desirable for industrial applications, likely
requiring implementation of some type of data signal lineariza-
tion or compensation to be able to advance towards commercial
devices.[193]
As a final remark of this section, a few recent works have been
conducted towards the exploration of hybrid architectures
combining CNTs and GSs mixed within polymer matri-
ces.[67,194,195] According to the initial studies on the piezoresistive
response of hybrid nanocomposites, the piezoresistive sensitivity
might be controlled by adjusting the CNT-to-GS ratio.[67,195] The
use of hybrid CNT-GS fillers may also bring about improvements
in the filler dispersion and load transfer from the matrix,
benefiting the mechanical properties of such nanocomposites.[196]
6. Piezoresistivity in Nanocomposites with
Aligned Fillers
Aligned carbon-based composites with anisotropic properties
are expected to render several advantages over those with
randomly oriented fillers.[197–199] One way of obtaining such
anisotropic properties is by forming directionally aligned
nanostructures within the polymer. In general, the formation
of aligned structures within polymer matrices can be obtained
during the composite formation (e.g., by infiltration, fiber
drawing, or application of electric or magnetic fields) or after the
incorporation of the fillers to the matrix (e.g., by stretching or
polymer compression).[198] Several attempts to align CB,[200,201]
CNTs,[202–209] and more recently, graphenic platelets,[210] have
been reported, although prefect alignment at the macro- and
nanoscales is yet a challenge. For the case of CNTs, excellent
reviews regarding the formation of CNT assemblies with
controlled orientation[211] and directional alignment of CNTs
within polymers can be found elsewhere.[198]
As a generally accepted outcome, the effective properties of the
composite material (e.g., mechanical, thermal, electrical) become
anisotropic upon alignment, that is, are strongly influenced by the
filler orientation. Since the intensity of the magnetic fields required
to align CNTs within polymers seems to be unpractically
high,[212,213] application of alternating current electric fields arises
as a more suitable option.[214–217] The electric field approach relies
on the electric dipole moment induced on the
CNTs due to the different electrical properties
of the CNTs and surrounding medium. The
induced moments promote CNT-to-CNT inter-
actions, where forces and torques tend to orient
and interconnect the CNTs.[218,219] This strategy
has been applied, for example, to align MWCNTs
in dissolved polysulfone,[217] as well as to align
SWCNTs in photo-polymerizable resins.[215,220]
Figure 17 shows examples of polymer compo-
sites with evident mesoscopically aligned struc-
tures conformed by CNTs, where the application
of an alternating current electric field was used.
As seen from both figures, preferential align-
ment of CNT bundles has been achieved at the Figure 17. CNT/polymer composites aligned by electric fields. a) MWCNT/polysulfone
meso-scale, but certain degree of misalignment composite,[217] b) SWCNT/acrylate composite.[220] The arrows indicate the direction of the
is yet visible. To date, alignment at the electric field.
Adv. Eng. Mater. 2018, 1701159 1701159 (15 of 23)
www.aem-journal.com
macroscale in polymer composites with evidence of perfect
individual CNT alignment at the nanoscale is yet a challenge.
The effect of filler alignment on the piezoresistive response of
carbon-based polymer composites has been scarcely explored,
with the majority of published studies focusing on
MWCNTs.[72,117,131,217,221] For aligned nanocomposites, the
piezoresistive response has been studied in the direction of
CNT alignment (axial) and perpendicular to it (transverse).
Table 2 summarizes the piezoresistive characteristics of different
types of aligned CNT arrays and CNT/polymer composites with
aligned CNTs, where the GFs were reported in at least two
directions, viz. axial, transverse (shortened as “Tran” in such a
Table), and/or randomly oriented. As seen from this table,
tensile testing is undoubtedly the most commonly implemented
technique to explore the alignment effect on the piezoresistive
response, although bending experiments have also been
reported. For the case of SWCNTs, most of the reports deal
with high SWCNT density systems without matrix, that is, in the
form of buckypaper (BP)[140,222] or arrays.[226,227] The highest GF
for a SWCNT aligned array (GF ¼ 75, under bending) was
reported by Cullinan and Culpepper,[227] where SWCNTs were
deposited on a bending beam and aligned by the application of
an alternating current electric field. This response evidences the
high intrinsic piezoresistive sensitivity of CNTs when they are
directly loaded (a challenging task). The rest of the GFs reported
for systems with aligned SWCNTs usually range from 1 to 10,
although some of them are not directly loaded but bonded to a
beam.[140,222,226] Recently, Sengezer et al.[220] aligned SWCNTs by
means of an electric field in a photopolymerizable blend of
acrylate resins, and investigated the piezoresistive response of
the nanocomposites along the axial (parallel to alignment) and
transverse (perpendicular to alignment) directions during
tensile loading. These authors reported higher piezoresistive
sensitivity for the transverse direction, see Table 2, which they
attributed to the interaction between misaligned and aligned
bridging structures, triggered by the Poisson’s contraction in the
transverse direction. For numerical comparison with existent
literature, it is important to keep in mind that the transverse
strain (calculated from the axial strain and Poisson’s ratio) was
used to calculate transverse GFs in Sengezer et al.[220] Another
factor to bear in mind for comparative purposes is that in their
work the load is always applied along the alignment direction
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Table 2. Summary of piezoresistive behavior of aligned CNT arrays and CNT/polymer composites with aligned CNTs.

Gage factor
Filler wt% Matrix (or substrate) Loading type Alignment method Axial Tran Random Ref.
 [222]
SWCNT BP PET T Mechanical draw 7.2 N/A 1.3
[220]
0.03 Acrylate T Electric field 2 10 N/A
 [223]
MWCNT BP PET B Magnetic field 2.1 4.5 2.3
Array Polyimide T Stretching 20 N/A 3 [224]

 [225]
Array Glass fiber/epoxy T CVD 8 1.3 N/A
[131]
1.0 PET T Extrusion 3.9 4.6 N/A
43 Epoxy T Stretching 0 0.25 N/A [117]

[217]
0.75 PSF T Electric field 1.5 N/A 0.7
[221]
5.0 PC T Injection molding 3.7 6.2 2.6

Acronyms: T ¼ Tensile, B ¼ Bending,  ¼ Bonded to a beam or to a larger tensile specimen, N/A ¼ Not available, BP ¼ Buckypaper, PC ¼ Polycarbonate, PSF ¼ Polysulfone,
PET ¼ Polyethylene terephthalate, CVD ¼ Chemical vapor deposition.

and the configuration of the longitudinal and transverse
electrodes are changed; different electrode configurations could
originate a different piezoresistive response per se. For the case
of MWCNTs, the piezoresistive response of several arrays of
aligned MWCNTs[223–225,228,229] and aligned MWCNT/polymer
composites[72,117,131,217,221] have also been reported. In general,
GFs of aligned MWCNTs and MWCNT composites range from
0.04 to 20. Thermosetting (epoxy) and thermoplastic (poly-
sulfone and polycarbonate) polymers have been used as host
matrices for aligned MWCNT networks with concentrations
ranging from 0.02[72] to 67 wt%.[117] As a general outcome, all the
studies confirm that the piezoresistive sensitivity becomes
anisotropic by aligning the CNTs, see Table 2. Some authors
have found higher GFs for the alignment direction than for the
randomly distributed case,[217,221,222,224] which may be explained
by the better strain transfer and increased contact resistance
along the aligned direction. To illustrate this, Figure 18 shows the
tensile piezoresistive response of a 200 μm thick MWCNT/PSF
composite film loaded along the alignment (axial) direction,
compared to the response of a film with randomly oriented
CNTs.[217]
Figure 18. Piezoresistive response of MWCNT/PSF composite films with
aligned and randomly oriented CNTs.
Only a few authors have compared the piezoresistive response
of CNT/polymer composites along both the alignment (axial) and
transverse directions,[117,131,220,221] generally finding that the GF in
the transverse direction is higher than that in the axial direction, see
Table 2. It is hypothesized that the reasons for this are the significant
change in CNT-to-CNT lateral contact and the breakdown of the few
conductive misaligned pathways bridging the aligned ones upon
strain application. However, more systematic studies are still needed
to better elucidate these mechanisms, and to investigate how several
factors such as the filler concentration, polymer rheology, alignment
degree, CNT length, functionalization, and loading configuration,
affect the electromechanical response of polymer nanocomposites
with aligned carbon fillers.
7. Concluding Remarks and Perspectives
The use of carbon-based polymer nanocomposites for develop-
ing smart piezoresistive materials which are capable of sensing
their own elastic strain, plasticity, and damage has been
reviewed. Upon reaching percolation, polymer composites
which are otherwise electrically insulating materials present
an electrical conductivity whose changes as a function of strain
can be employed for self-sensing of strain and damage. Carbon
materials such as graphite, carbon black, carbon nanotubes, and
lately few- and multi-layer graphene are ideal candidates to act as
polymer fillers to such an aim. The coupled electromechanical
response of such polymer nanocomposites depend on several
factors related to the filler (nanostructure architecture, dimen-
sions, number of layers, functional groups), to the polymer
matrix (dielectric constant, processing viscosity) and to the filler
dispersion within the matrix. From the geometrical aspects of
the filler, the excluded volume is one of the key parameters that
influences the piezoresistive sensitivity of polymer composites.
It is expected that loosely connected networks with low filler
concentrations, high excluded volume and high number of filler-
to-filler contact points render higher piezoresistive sensitivity.
Creation and destruction of conductive pathways upon loading,
change in contact resistance and change in tunneling resistance
are believed to be the main mechanisms governing the

Adv. Eng. Mater. 2018, 1701159 1701159 (16 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

piezoresistive response of these smart materials. The use of
piezoimpedance and piezocapacitance for strain and damage
sensing in this kind of polymer composites is also an interesting
approach that deserves further exploration. The current
applications of thermosetting and thermoplastic piezoresistive
materials are expected to provide the automotive, aerospace,
transportation and energy industries with the next generation
of smart materials. Especially for thermosetting polymers, the
applications of these nanocomposites are rapidly evolving
towards self-sensing of strain and damage in fiber-reinforced
multiscale hierarchical composites, for virtually every industrial
application of composites which demand high load bearing
capacity and self-sensing. The applications of carbon filled
elastomeric materials, on the other hand, focus more on sensors
for human motion, soft skins, smart wearable sensors and
robotics, whose main requisite is large deformations. However,
aspects such as nonlinearity, cycling reproducibility, undesired
viscoelastic response, and hysteresis, are practical issues that
need to be resolved towards further development of commercial
devices based on these materials.
Because of their elongated geometry, carbon nanotubes stand
today as ideal candidates for fillers in this kind of piezoresistive
materials. Controlling the morphology an orientation of the
carbon nanotube networks, for example, by aligning them in a
preferential direction, may render tailored piezoresistive sensitivi-
ties. However, the higher surface area of two-dimensional
graphenic platelets and the possibility of their top-down synthesis
in large scales are features which make them also very attractive for
piezoresistive applications. As a future trend, hybrid combinations
of one dimensional carbon nanotubes and two-dimensional
graphenic platelets may become optimal topological combinations
for piezoresistive strain sensing applications, but the optimization
of the hybrid architecture is yet to come.
Appendix A: Compendium of Piezoresistive
Behavior of Carbon Filled Polymer Composites
The main piezoresistive features of about 90 works are collected
in the parameters listed in Table A1. Such a table is organized
by filler type and includes thermosetting (TS), thermoplastic
(TP) and elastomeric (ES) matrices, as well as the correspond-
ing filler concentration (wt%) and loading type (tension,
compression or bending). The main features of the piezor-
esistive response of the material are captured by four
parameters, referring to its linear (L) or nonlinear (NL)
character, maximum strain sensing capability (emax), extend
of linear piezoresistivity (eL) and piezoresistive sensitivity (GF).
The difference between eL and emax provides an indication of
nonlinear piezoresistivity. All concentrations are reported in
weight (wt%), so all values originally reported in volume
concentration (vol%) were converted to weight (and thus,

Table A1. Summary of piezoresistive behavior and gage factors of polymer composites filled with carbon nanostructures.

Piezoresistivity
Filler wt% Matrix (or substrate) Loading type L/NL emax (%) eL (%) GF Ref.
[71]
Carbon black 0.5 Epoxy T NL 6 1 6
[72]
12 Epoxy T NL 4 1 27.7
[73]
50.8 HDPE C L 50 50 3.2
[74]
13.4 PP T NL 3 1.5 6.6
[75]
5.2 PP T L 3 3 30
[76]
9.2 HDPE T NL 25 15 10
[77]
12.6 HDPE T NL 90 35 4.5
[78]
3 PVDF T NL 9 5 3
[79]
33 Natural rubber T NL 90 7 17
[80]
30.6 Natural rubber T NL 250 100 12
[82]
10 Polyisoprene T NL 40 10 15
[83]
9.0 Silicone rubber C NL 25 2.5 3.6
50 Silicone rubber C NL 25 2.5 4.4 [83]

7.0 Silicone rubber C NL 7 4 6.7 [84]

[73]
50.2 Silicone rubber C NL 50 13 4.2
35 Natural rubber C L 16.7 16.7 2.7 [85]

 [96]
Graphite 16.7 Epoxy B N/A N/A N/A 22
 [95]
18 Polyimide B NL N/A N/A 16.8
[91]
12.5 PP T NL 0.6 0.3 135
[92]
37.5 HDPE C NL 35 5 14

(Continued)

Adv. Eng. Mater. 2018, 1701159 1701159 (17 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Table A1. (Continued)

Piezoresistivity
Filler wt% Matrix (or substrate) Loading type L/NL emax (%) eL (%) GF Ref.
[6]
27.6 HDPE C NL 10 7 7
[93]
49.4 Silicone rubber C NL 15 3 44
3.1 Silicone rubber C NL 0.5 0.3 16.7 [94]


SWCNT BP PDMS T NL 280 40 0.82 (0 < e < 0.4) [20]

[232]
BP Epoxy T NL 0.95 0.6 2.2
[232]
BP R-Epoxy T NL 15 3 0.60
 [140]
BP PET T L 2.0 N/A 1.0

BP PDMS T NL 30 N/A 5.0 (0.2< e < 0.3) [140]

[233]
BP N/A T L 1.0 1.0 1.25
[222]
BP PET T L 0.65 N/A 1.30 to 7.19
 [113]
0.2 Epoxy B NL 3.0 1.0 3.6
 [122]
0.5 PMMA B L 0.13 0.13 5.2
 [122]
10 PMMA B L 0.13 0.13 0.95
[123]
2.0 PVDF B L 0.25 0.25 6.2
 [124]
2.0 PVDF B L 0.6 0.6 4.43
8.0 TPU T NL 60 20 0.05 to 0.10 (e < 0.2) [137]

[145]
1.0 SBS T L 5 5 3 to 4.5
[232]
MWCNT BP Epoxy T NL 1.07 0.6 0.85
[232]
BP R-Epoxy T NL 30 7 0.70
 [135]
BP TPU T NL 400 0 4 to 69
[234]
BP TPU T NL 12 0 5 to 44
[71]
0.1 Epoxy T NL 6.0 1.5 3.4 to 4.3
 [114]
5.0 Epoxy T (C) L 0.6 0.6 3.8 (4.9)
[114]
1.0 Epoxy T (C) NL 0.6 0.1 22.4 (7.1)
0.2 Epoxy C NL 12 2 4.2 [115]

[116]
0.3 Epoxy T NL 7.0 1.9 1.42
[117]
53 Epoxy T L 1.0 1.0 1.0 to 1.5
[110]
0.3 Epoxy T NL 3.5 2.0 0.43
0.3 Epoxy B NL 3.5 <0.3 <0.1 [110]

[72]
0.3 Epoxy T NL 4.0 1.0 4.1
[112]
0.3 Vinyl ester T L 1.3 1.3 2.60
0.3 Vinyl ester C NL 20 1.0 0.91 (EL), 9.37 (PL) [112]

[128]
5.0 PC T L 1.7 1.0 7.0
1.0 PEO T NL 1.0 0.8 3.7 [129]

[32]
1.0 PMMA T NL 1.0 0.2 15.3
[32]
3.0 PMMA T NL 1.0 0.2 4.6
[130]
1.0 PSF T NL 5.0 1.0 0.73
[131]
1.0 PET T NL 5.0 0.5 0.71 (EL), 3.9 (PL)
 [126]
0.75 PVDF B L 0.3 0.3 3.91
[74]
7.6 PP T NL 3 1.0 1.3
[33]
4.0 PP T NL 9.7 1.0 3.56 (EL), 28.9 (PL)
[33]
8.0 PP T NL 8.0 1.0 1.23 (EL), 23.9 (PL)
[139]
1.4 PDMS T NL 50 20 1 to 2
2.3 TPU T NL 1000 0 5 (e ¼ 2.5) [136]

(Continued)

Adv. Eng. Mater. 2018, 1701159 1701159 (18 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.aem-journal.com

Table A1. (Continued)

Piezoresistivity
Filler wt% Matrix (or substrate) Loading type L/NL emax (%) eL (%) GF Ref.
[138]
2.0 TPU T NL 30 0 2 to 6
[134]
6.0 TPU T NL 100 0 0.09 to 2.64
[145]
1.0 SBS T L 5 5 2 to 8
[146]
4.0 SBS T L 50 10 3 to 120
 [146]
4.0 SBS B L 0.065 0.065 40 to 100
[142]
3.0 Silicone rubber T NL 10 0 2.0 to 9.46
[153]
4.0 TPU T NL 100 0 21.9 to 126

2.0 SEBS T NL 80 N/A 1.2 to 2.2 [143]


4.0 SEBS T NL 80 N/A 1.5 to 2.7 [143]

[189]
Graphenic sheets GP PDMS T NL 100 10 2
 [97]
GP PDMS T L 1.6 1.6 546
 [171]
GP PMMA B L 0.8 0.8 53
[235]
GP PDMS T NL 26 6 34.3
2.2 PVDF T L 1.2 1.2 12.1 [184]


7.0 PVDF T L N/A N/A 5 [184]

 [185]
5.0 PVDF B L 0.15 0.15 4 to 11
 [179]
3.0 Epoxy B L 0.1 0.1 11.4
2.0 Epoxy B NL 2.6 0.6 56.7 [191]

 [180]
1.0 Epoxy B NL 2.35 1 63.5
 [180]
2.0 Epoxy B NL 2.35 1 23.4
[183]
2 Epoxy T NL 2 1 12.8
[172]
2.7 Epoxy T L N/A N/A 45
[173]
2.0 Epoxy T NL 3 0.5 24
[173]
5.0 Epoxy T NL 3 0.5 12
 [181]
1.0 Vinyl ester B NL 1.5 0.3 33
[174]
0.5 PDMS T NL 95 45 0.4
 [178]
15.6 PDMS T L 2.0 2.0 233
 [178]
31 PDMS T L 2.0 2.0 3.9
 [182]
20 PDMS B NL 9.0 6 30
30 PDMS B NL 9.0 6 10 [182]

11 Polysilicone T(C) NL 10% (100%) <1% 500 (40) [175]

[176]
1.0 TPU C NL 55 12 6.2
2.0 TPU C NL 90 60 2.45 [177]

[61]
19 EB T NL 20 10 53
[61]
21 EB T NL 20 10 33

Acronyms: T ¼ Tensile, C ¼ Compressive, B ¼ Bending,  ¼ Bonded to a beam or to a larger tensile specimen, L ¼ Linear, NL ¼ Nonlinear, emax ¼ Maximum strain registered
by the electrical signal, eL ¼ Maximum strain corresponding to approximately linear piezoresistivity, EL ¼ Elastic region, PL ¼ Plastic region, N/A ¼ Not available,
BP ¼ Buckypaper, GP ¼ Graphene-paper, R-Epoxy ¼ Rubber-modified epoxy, PP ¼ Polypropylene, PC ¼ Polycarbonate, HDPE ¼ high denstity polyethylene, PET ¼
polyethylene terephthalate, PSF ¼ Polysulfone, PEO ¼ Polyethylene oxide, PMMA ¼ Polymethyl methacrylate, PDMS ¼ Polydimethylsiloxane, TPU ¼ Thermoplastic
polyurethane, PVDF ¼ Polyvinylidene fluoride, EPDM ¼ Ethylene propylene diene rubber, EB ¼ ethylene-butene, SBS ¼ styrene-butadiene-styrene, SEBS ¼ styrene-
ethylene-butylene-styrene.

represent our best estimate). For conversions from vol% to wt%
the densities of the carbon nanostructures were taken as
1800 kg m3 (carbon black),[80] 2200 kg m3 (graphite),[230]
2100 kg m3 (MWCNT),[231] and 1870 kg m3 (SWCNT).[231]
For the majority of the cases, the numerical value of the gage
factor was directly provided by the original reference. In the
cases that such a number was not directly provided, but the
required piezoresistive curve was published, the available data
was digitalized and the GFs reported in Table A1 were
estimated.

Adv. Eng. Mater. 2018, 1701159 1701159 (19 of 23) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Acknowledgements
The results of the authors collected in this review were possible due to the
sponsorship of CONACYT (Mexico) through many years of research. In
particular, grants No. 79609, 220513, 188089 (CIAM), directed by Dr. FA
as well as grant No. 235905 directed by Dr. AO are especially
acknowledged. Technical assistance of Dr. Alejandro May at CICY as
well as the valuable input of the group of students of Dr. Avil
es through
many years of research are also highly valued. The original
Figure 17b (unpublished) was kindly provided by Dr. Engin Sengezer
working with Dr. Gary Seidel at Virginia Tech, and we are very thankful to
both of them for that. Artwork by Omar Rodríguez and Rosario May is
appreciated.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
carbon nanostructures, nanocomposites, piezoresistivity, self-sensing
Received: December 23, 2017
Revised: February 21, 2018
Published online:
[1] A. S. Kobayashi, Handbook on Experimental Mechanics, VCH
Publishers, New York, USA 1993, p. 1074.
[2] C. Liu, Foundations of MEMS, Prentice Hall, Uppers Saddle River,
USA 2006, p. 530.
[3] W. Thomson, Philos. Trans. R. Soc. London 1856, 146, 649.
[4] X. Wang, D. D. L. Chung, Carbon 1997, 35, 1649.
[5] D. D. L. Chung, Adv. Eng. Mater. 2000, 2, 788.
[6] J. Lu, X. Chen, W. Lu, G. Chen, Eur. Polym. J. 2006, 42, 1015.
[7] S. Qu, S.-C. Wong, Compos. Sci. Technol. 2007, 67, 231.
[8] L. Flandin, A. Hiltner, E. Baer, Polymer 2001, 42, 827.
[9] P. Murugaraj, D. E. Mainwaring, N. Mora-Huertas, Compos. Sci.
Technol. 2009, 69, 2454.
[10] C. Li, E. T. Thostenson, T.-W. Chou, Compos. Sci. Technol. 2008, 68,
1227.
[11] K. S. Karimov, F. A. Khalid, M. T. S. Chani, Measurement 2012, 45,
918.
[12] J. W. Jo, J. U. Lee, W. H. Jo, Polym. Int. 2015, 64, 1676.
[13] J. Zhao, G. Wang, R. Yang, X. Lu, M. Cheng, C. He, G. Xie, J. Meng,
D. Shi, G. Zhang, ACS Nano 2015, 9, 1622.
[14] P. Greil, Adv. Eng. Mater. 2015, 17, 124.
[15] M. Amjadi, K.-U. Kyung, I. Park, M. Sitti, Adv. Funct. Mater. 2016, 26,
1678.
[16] A. J. C. Hierold, C. Stampfer, T. Helbling, Sens. Actuators A 2007,
136, 51.
[17] M. F. L. De Volder, S. H. Tawfick, R. H. Baughman, J. Hart, Science
2013, 339, 535.
[18] L. M. Castano, A. B. Flatau, Smart Mater. Struct. 2014, 23, 053001.
[19] P. Polygerinos, N. Correll, S. A. Morin, B. Mosadegh, C. D. Onal,
K. Petersen, M. Cianchetti, M. T. Tolley, R. F. Shepherd, Adv. Eng.
Mater. 2017 2017, 19, 1700016.
[20] T. Yamada, Y. Hayamizu, Y. Yamamoto, Y. Yomogida, A. Izadi-
Najafabadi, D. N. Futaba, K. Hata, Nat. Nanotechnol. 2011, 6, 296.
[21] N. Wiegand, E. Mäder, Adv. Eng. Mater. 2016, 18, 376.
[22] J. J. Ku-Herrera, A. May-Pat, F. Avil
es, Adv. Eng. Mater. 2016, 18, 963.
Adv. Eng. Mater. 2018, 1701159 1701159 (20 of 23)
www.aem-journal.com
[23] M.-Y. Cheng, C.-M. Tsao, Y.-Z. Lai, Y.-J. Yang, Sens. Actuators A
2011, 166, 226.
[24] N. Li, S. Oida, G. S. Tulevski, S.-J. Han, J. B. Hannon, D. K. Sadana,
T.-C. Chen, Nat. Commun. 2013, 4, 2294.
[25] A. I. Oliva-Avil
es, F. Avil
es, G. D. Seidel, V. Sosa, Composites Part B
2013, 47, 200.
[26] G. E. Pike, C. H. Seager, Phys. Rev. B 1974, 10, 1421.
[27] W. Bauhofer, J. Z. Kovacs, Compos. Sci. Technol. 2009, 69, 1486.
[28] J. K. W. Sandler, J. E. Kirk, I. A. Kinloch, M. S. P. Shaffer, A. H. Windle,
Polymer 2003, 44, 893.
[29] N. Hu, Z. Masuda, C. Yan, G. Yamamoto, H. Fukunaga, T. Hashida,
Nanotechnology 2008, 19, 215701.
[30] B. Fiedler, F. H. Gojny, M. H. G. Whichmann, W. Bauhofer,
K. Schulte, Ann. Chim. Sci. Mater. 2004, 29, 81.
[31] J. J. Ku-Herrera, O. F. Pacheco-Salazar, C. R. Ríos-Soberanis,
G. Domínguez-Rodríguez, F. Avil
es, Sensors 2016, 16, 400.
[32] G. T. Pham, Y-B. Park, Z. Liang, C. Zhang, B. Wang, Composites Part
B 2008, 39, 209.
[33] O. Zetina-Hernández, S. Duarte-Aranda, A. May-Pat, G. Canch
e-
Escamilla, J. Uribe-Calderon, P. I. Gonzalez-Chi, F. Avil
es, J. Mater.
Sci. 2013, 48, 7587.
[34] C. Stampfer, A. Jungen, R. Linderman, D. Obergfell, S. Roth,
C. Hierold, Nano Lett. 2006, 6, 1449.
[35] X.-W. Fu, Z.-M. Liao, J.-X. Zhou, Y.-B. Zhou, H.-C. Wu, R. Zhang,
G. Jing, J. Xu, X. Wu, W. Guo, D. Yu, Appl. Phys. Lett. 2011, 99,
213107.
[36] W. Obitayo, T. Liu, J. Sens. 2012, 2012, 652438.
[37] X. Ren, G. D. Seidel, J. Intell. Mater. Syst. Struct. 2013, 24, 1459.
[38] L. Flandin, A. Chang, S. Nazarenko, A. Hiltner, E. Baer, J. Appl.
Polym. Sci. 2000, 76, 894.
[39] T. C. Theodosiou, D. A. Saravanos, Compos. Sci. Technol. 2010, 70,
1312.
[40] M. Mohiuddin, S. V. Hoa, Compos. Sci. Technol. 2013, 79, 42.
[41] X. Ren, A. K. Chaurasia, A. I. Oliva-Avil
es, J. J. Ku-Herrera,
G. D. Seidel, F. Avil
es, Smart Mater. Struct. 2015, 24, 065031.
[42] W. Fang, S. N. Leung, J. Appl. Phys. 2015, 118, 044907.
[43] J. G. Simmons, J. Appl. Phys. 1963, 34, 2581.
[44] N. Hu, Y. Karube, C. Yan, Z. Masuda, H. Fukunaga, Acta Mater.
2008, 56, 2929.
[45] A. K. Chaurasia, G. D. Seidel, J. Intell. Mater. Syst. Struct. 2014, 25, 2141.
[46] C. Li, E. T. Thostenson, T-W. Chou, Appl. Phys. Lett. 2007, 91,
223114.
[47] B. M. Lee, K. J. Loh, J. Mater. Sci. 2015, 50, 2973.
[48] F. Panozzo, M. Zappalorto, M. Quaresimin, Composites Part B 2017,
109, 53.
[49] T. Tallman, K. W. Wang, Appl. Phys. Lett. 2013, 102, 011909.
[50] B. M. Lee, K. J. Loh, Y.-S. Yang, Comput. Mech. 2017, 60, 39.
[51] Z. Wang, X. Ye, Nanotechnology 2013, 24, 265704.
[52] S. Xu, O. Rezvanian, M. A. Zikry, Smart Mater. Struct. 2013, 22,
055032.
[53] R. Rahman, P. Servati, Nanotechnology 2012, 23, 055703.
[54] S. Gong, Z. H. Zhu, S. A. Meguid, Polymer 2014, 55, 5488.
[55] F. Avil
es, A. May-Pat, G. Canch
e-Escamilla, O. Rodríguez-Uicab, J. J. Ku-
Herrera, S. Duarte-Aranda, J. Uribe-Calderon, P. I. Gonzalez-Chi,
L. Arronche, V. La Saponara, J. Intell. Mater. Syst. Struct. 2016, 27, 92.
[56] R. Socher, B. Krause, M. T. Müller, R. Boldt, P. Pötschke, Polymer
2012, 53, 495.
[57] C. Cattin, P. Hubert, ACS Appl. Mater. Interfaces 2014, 6, 1804.
[58] R. Hashemi, G. J. Weng, Carbon 2016, 96, 474.
[59] A. Celzard, E. McRae, C. Deleuze, M. Dufort, G. Furdin,
J. F. Marech
e, Phys. Rev. B 1996, 53, 6209.
[60] Y. Sun, H.-D. Bao, Z.-X. Guo, J. Yu, Macromolecules 2009, 42, 459.
[61] H. Palza, C. Garzon, M. Rojas, Polym. Int. 2016, 65, 1441.
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
[62] I. Balberg, C. H. Anderson, S. Alexander, N. Wagner, Phys. Rev. B
1984, 30, 3933.
[63] A. P. Philipse, Langmuir 1996, 12, 1127.
[64] E. J. Garboczi, K. A. Snyder, J. F. Douglas, M. F. Thorpe, Phys. Rev. E
1995, 52, 819.
[65] S. R. Williams, A. P. Philipse, Phys. Rev. E 2003, 67, 051301.
[66] E. Andreoli, K.-S. Liao, A. Cricini, X. Zhang, R. Soffiatti, H. J. Byrne,
S. A. Curran, Thin Solid Films 2014, 550, 558.
[67] H. Aguilar-Bolados, M. Yazdani-Pedram, A. Contreras-Cid,
M. A. L
opez-Manchado, A. May-Pat, F. Avil
es, Composites Part B
2017, 109, 147.
[68] A. Wouterse, S. Luding, A. P. Philipse, Granular Matter 2009, 11,
169.
[69] A. Proper, W. Zhang, S. Bartolucci, A. A. Oberai, N. Koratkar,
Nanosci. Nanotechnol. Lett. 2009, 1, 3.
[70] M. Mohiuddin, S. V. Hoa, Compos. Sci. Technol. 2011, 72, 21.
[71] M. H. G. Wichmann, S. T. Buschhorn, J. Gehrmann, K. Schulte,
Phys. Rev. B 2009, 80, 245437.
[72] H. Meeuw, C. Viets, W. V. Liebig, K. Schulte, B. Fiedler, Eur. Polym.
J. 2016, 85, 198.
[73] P. Wang, T. Ding, J. Appl. Polym. Sci. 2010, 116, 2035.
[74] J. Zhao, K. Dai, C. Liu, G. Zheng, B. Wang, C. Liu, J. Chen, C. Shen,
Composites Part A 2013, 48, 129.
[75] Y. Qu, K. Dai, J. Zhao, G. Zheng, C. Liu, J. Chen, C. Shen, Colloid.
Polym. Sci. 2014, 292, 945.
[76] S. Zheng, J. Deng, L. Yang, D. Ren, S. Huang, W. Yang, Z. Liu,
M. Yang, Compos. Sci. Technol. 2014, 97, 34.
[77] S. Zheng, J. Deng, L. Yang, D. Ren, W. Yang, Z. Liu, M. Yang, RSC
Adv. 2015, 5, 31074.
[78] K. Ke, P. Pötschke, N. Wiegand, B. Krause, B. Voit, ACS Appl. Mater.
Interfaces 2016, 8, 14190.
[79] D. Bulgin, Rubber Chem. Technol. 1946, 19, 667.
[80] P. E. Wack, R. L. Anthony, E. Guth, J. Appl. Phys. 1947, 18, 456.
[81] A. Voet, F. R. Cook, Rubber Chem. Technol. 1968, 41, 1207.
[82] M. Knite, V. Teteris, A. Kiploka, J. Kaupuzs, Sens. Actuators A 2004,
110, 142.
[83] W. E. Mahmoud, A. M. Y. El-Lawindy, M. H. El Eraki, H. H. Hassan,
Sens. Actuators A 2007, 136, 229.
[84] L. Wang, T. Ding, P. Wang, Compos. Sci. Technol. 2008, 65, 3448.
[85] Q. He, T. Yuan, X. Zhang, S. Guo, J. Liu, J. Liu, X. Liu, L. Sun, S. Wei,
Z. Guo, Mater. Res. Express 2014, 1, 035029.
[86] I. Estrada-Moreno, A. Díaz-Diaz, M. Mendoza-Duarte, R. Ibarra-
G
omez, Macromol. Symp. 2009, 283–584, 361.
[87] M. Melnykowycz, B. Koll, D. Scharf, F. Clemens, Sensors 2017, 14,
1278.
[88] L. Wang, F. Ma, Q. Shi, H. Liu, X. Wang, Sens. Actuators A 2011, 165, 207.
[89] L. Wang, T. Ding, P. Wang, Carbon 2009, 47, 3151.
[90] D. S. A. De Focatiis, D. Hull, A. Sánchez-Valencia, Plast. Rubber
Compos. 2012, 41, 7.
[91] V. G. Shevchenko, A. T. Ponomarenko, C. Klason, Smart. Mater.
Struct. 1995, 4, 31.
[92] Q. Zheng, Y. Song, X.-S. Yi, J. Mater. Sci. Lett. 1999, 18, 35.
[93] D. T. Beruto, M. Capurro, G. Marro, Sens. Actuators A 2005, 117,
301.
[94] L. Chen, G. Chen, L. Lu, Adv. Funct. Mater. 2007, 17, 898.
[95] A. B. Frazier, K. G. Allen, J. Appl. Phys. 1993, 73, 4428.
[96] N. Serra, T. Maeder, P. Ryser, Sens. Actuators A 2012, 186, 198.
[97] H. Zhao, J. Bai, ACS Appl. Mater. Interfaces 2015, 7, 9652.
[98] Y. He, W. Li, G. Yang, H. Liu, J. Lu, T. Zheng, X. Li, Materials 2017,
10, 684.
[99] A. E. Job, F. A. Oliveira, N. Alves, J. A. Giacometti, L. H. C. Mattoso,
Synth. Met. 2003, 135–136, 99.
[100] J. F. Zhou, Y. H. Song, Q. Zheng, Q. Wu, M. Q. Zhang, Carbon 2008,
46, 679.
Adv. Eng. Mater. 2018, 1701159 1701159 (21 of 23)
www.aem-journal.com
[101] L. Wang, T. Ding, P. Wang, Sens. Actuators A 2007, 135, 587.
[102] L. Wang, T. Ding, P. Wang, J. Polym. Sci. B Polym. Phys. 2008, 46,
1050.
[103] S.-L. Wang, P. Wang, T.-H. Ding, J. Appl. Polym. Sci. 2009, 113, 337.
[104] J. Zavickis, M. Knite, G. Podins, A. Linarts, R. Orlovs, Sens. Actuators
A 2011, 171, 38.
[105] J. Zavickis, M. Knite, K. Ozols, G. Malefan, Mater. Sci. Eng. C 2011,
31, 472.
[106] F. Carmona, R. Canet, P. Delhaes, J. Appl. Phys. 1987, 61, 2550.
[107] G. Chen, J. Lu, W. Lu, D. Wu, C. Wu, Polym. Int. 2005, 54, 1689.
[108] L. Wang, Sens. Actuators A 2016, 252, 89.
[109] P. Dharap, Z. Li, S. Nagarajaiah, E. V. Barrera, Nanotechnology 2004,
15, 379.
[110] L. Vertuccio, L. Guadagno, G. Spinelli, P. Lamberti, V. Tucci,
S. Russo, Composites Part B 2016, 107, 192.
[111] C. Lozano-P
erez, J. V. Cauich-Rodríguez, F. Avil
es, J. Appl. Polym.
Sci. 2017, 134, 44448.
[112] J. J. Ku-Herrera, F. Avil
es, G. D. Seidel, Smart Mater. Struct. 2013, 22,
085003.
[113] A. de la Vega, I. A. Kinloch, R. J. Young, W. Bauhofer, K. Schulte,
Compos. Sci. Technol. 2011, 71, 160.
[114] G. Yin, N. Hu, Y. Karube, Y. Liu, Y. Li, H. Fukunaga, J. Compos.
Mater. 2011, 45, 1315.
[115] N. J. Heeder, A. Shukla, V. Chalivendra, S. Yang, K. Park, Exp. Mech.
2012, 52, 315.
[116] V. K. Vadlamani, V. B. Chalivendra, A. Shukla, S. Yang, Polym.
Compos. 2012, 33, 1809.
[117] M. Mecklenburg, D. Mizushima, N. Ohtake, W. Bauhofer, B. Fiedler,
K. Schulte, Carbon 2015, 91, 275.
[118] K. Aly, A. Li, P. D. Bradford, Composites Part B 2017, 116, 459.
[119] K. Aly, A. Li, P. D. Bradford, J. Compos. Mater. 2018, https://doi.org/
10.1177/0021998317754128.
[120] B. Pinto, S. Kern, J. J. Ku-Herrera, J. Yasui, V. L. Saponara, K. J. Loh,
J. Phys. Conf. Ser. 2015, 628, 012098.
[121] J. J. Ku-Herrera, F. Avil
es, Carbon 2012, 50, 2592.
[122] I. Kang, M. Schulz, J. H. Kim, V. Shanov, D. Shi, Smart Mater. Struct.
2006, 15, 737.
[123] A. Ferreira, J. G. Rocha, A. Ans
on-Casaos, M. T. Martínez, F. Vaz,
S. Lanceros-Mendez, Sens. Actuators A 2012, 178, 10.
[124] A. Ferreira, S. Lanceros-Mendez, Composites Part B 2016, 96, 242.
[125] A. Ferreira, V. Correia, E. Mendes, C. Lopes, J. F. Vilela Vaz,
S. Lanceros-Mendez, IEEE Sens. J. 2017, 17, 2182.
[126] A. Ferreira, M. T. Martínez, A. Ans
on-Casaos, L. E. G
omez-Pineda,
F. Vaz, S. Lanceros-Mendez, Carbon 2013, 61, 568.
[127] S. Vidhate, J. Chung, V. Vaidyanathan, N. D’Souza, Mater. Lett.
2009, 63, 1771.
[128] W. Zhang, J. Suhr, N. Koratkar, J. Nanosci. Nanotechnol. 2006, 6, 960.
[129] M. Park, H. Kim, J. P. Youngblood, Nanotechnology 2008, 19,
055705.
[130] J. R. Bautista-Quijano, F. Avil
es, J. O. Aguilar, A. Tapia, Sens.
Actuators A 2010, 159, 135.
[131] F. Nanni, B. L. Mayoral, F. Madau, G. Montesperelli, T. McNally,
Compos. Sci. Technol. 2012, 72, 1140.
[132] R. Zhang, M. Baxendale, T. Peijs, Phys. Rev. B 2007, 76, 195433.
[133] Z.-J. Wang, D.-J. Kwon, G.-Y. Gu, H.-S. Kim, D.-S. Kim, C.-S. Lee,
K. L. DeVries, J.-M. Park, Compos. Sci. Technol. 2013, 81, 69.
[134] J. R. Bautista-Quijano, F. Avil
es, J. V. Cauich-Rodriguez, J. Appl.
Polym. Sci. 2013, 130, 375.
[135] P. Slobodian, P. Riha, P. Saha, Carbon 2012, 50, 3446.
[136] Q. Fan, Z. Qin, S. Gao, Y. Wu, J. Pionteck, E. Mäder, M. Zhu, Carbon
2012, 50, 4085.
[137] J. R. Bautista-Quijano, F. Avil
es, J. V. Cauich-Rodríguez,
R. Schönfelder, A. Bachmatiuk, T. Gemming, M. H. Rümmeli,
Synth. Met. 2013, 185–186, 96.
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
[138] R. Zhang, H. Deng, R. Valenca, J. Jin, Q. Fu, E. Bilotti, T. Peijs,
Compos. Sci. Technol. 2013, 74, 1.
[139] J.-B. Lee, D.-Y. Khang, Compos. Sci. Technol. 2012, 72, 1257.
[140] S. Luo, T. Liu, Carbon 2013, 59, 315.
[141] J. Hwang, J. Jang, K. Hong, K. N. Kim, J. H. Han, K. Shin, C. E. Park,
Carbon 2011, 49, 106.
[142] L. Wang, L. Cheng, Carbon 2014, 71, 319.
[143] B. F. Goncalves, P. Costa, J. Oliveira, S. Ribeiro, V. Correia,
G. Botelho, S. Lanceros-Mendez, J. Polym. Sci. Polym. Phys. 2016,
54, 2092.
[144] H. F. Castro, V. Correia, N. Pereira, P. Costab, J. Oliveira,
S. Lanceros-M
endez, Addit. Manuf. 2018, 20, 119.
[145] P. Costa, J. Silva, A. Ans
on-Casaos, M. T. Martinez, M. J. Abad,
J. Viana, S. Lanceros-Mendez, Composites Part B 2014, 61, 136.
[146] P. Costa, A. Ferreira, V. Sencadas, J. C. Viana, S. Lanceros-M
endez,
Sens. Actuators A 2013, 201, 458.
[147] L. Wang, Z. Wang, Y. Wang, X. Wang, H. Wang, G. Lu, D. Zhao,
Z. Li, J. Appl. Polym. Sci. 2016, 133, 42945.
[148] P. Costa, C. Silvia, J. C. Viana, S. Lanceros-Mendez, Composites Part
B 2014, 57, 242.
[149] I. Kang, Md. A. Khaleque, Y. Yoo, P. J. Yoon, S-Y. Kim, K. T. Lim,
Composites Part A 2011, 42, 623.
[150] B. Fernández-d’Arlas, U. Khan, L. Rueda, J. N. Coleman,
I. Mondragon, M. A. Corcuera, A. Eceiza, Compos. Sci. Technol.
2011, 71, 1030.
[151] L. Wang, C. Xy, Y. Li, Sens. Actuators A 2013, 189, 45.
[152] Y. Kazemi, A. R. Kakroodi, A. Ameli, T. Filleter, C. B. Park, Mater.
Chem. C 2018, 6, 350.
[153] C. Lozano-P
erez, J. V. Cauich-Rodríguez, F. Avil
es, Compos. Sci.
Technol. 2016, 128, 25.
[154] T. M. Madkour, F. M. Hagag, W. Mamdouh, R. A. Azzam, Polymer
2012, 53, 5788.
[155] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang,
S. V. Dubonos, I. V. Grigorieva, A. A. Firsov, Science 2004, 306,
666.
[156] F. Schedin, A. K. Geim, S. V. Morozov, E. W. Hill, P. Blake,
M. I. Katsnelson, K. S. Novoselov, Nat. Mater. 2007, 6, 652.
[157] H. A. Becerril, J. Mao, Z. Liu, R. M. Stoltenberg, Z. Bao, Y. Chen, ACS
Nano 2008, 2, 463.
[158] Y. Lee, S. Bae, H. Jang, S. Jang, S. E. Zhu, S. H. Sim, Y. I. Song,
B. H. Hong, J.-H. Ahn, Nano Lett. 2010, 10, 490.
[159] H. Aoki, M. S. Dresselhaus, Physics of Graphene, Springer, New
York, USA 2014.
[160] A. Bianco, H.-M. Cheng, T. Enoki, Y. Gogotsi, R. H. Hurt,
N. Koratkar, T. Kyotani, M. Monthioux, C. R. Park, J. M. D. Tascon,
J. Zhang, Carbon 2013, 65, 1.
[161] D. Wei, Y. Liu, Adv. Mater. 2010, 22, 3225.
[162] Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts, R. S. Ruoff, Adv.
Mater. 2010, 22, 3906.
[163] P. Wick, A. E. Louw-Gaume, M. Kucki, H. F. Krug, K. Kostarelos,
B. Fadeel, K. A. Dawson, A. Salvati, E. Vázquez, L. Ballerini,
M. Tretiach, F. Benfenati, E. Flahaut, L. Gauthier, M. Prato,
A. Bianco, Angew. Chem. Int. Ed. 2014, 53, 7714.
[164] M. Huang, T. A. Pascal, H. Kim, W. A. Goddard, J. R. Greer, Nano
Lett. 2011, 11, 1241.
[165] A. D. Smith, F. Niklaus, A. Paussa, S. Vaziri, A. C. Fischer,
€
M. Sterner, F. Forsberg, A. Delin, D. Esseni, P. Palestri, M. Ostling,
M. C. Lemme, Nano Lett. 2013, 13, 3237.
[166] J. Zhao, C. He, R. Yang, Z. Shi, M. Cheng, W. Yang, G. Xie, D. Wang,
D. Shi, G. Zhang, Appl. Phys. Lett. 2012, 101, 063112.
[167] X. Zheng, X. Chen, J.-K. Kim, D.-W. Lee, X. Li, J. Micro/Nanolithogr.
MEMS, MOEMS 2013, 12, 013009.
[168] S.-E. Zhu, M. K. Ghatkesar, C. Zhang, G. C. A. M. Janssen, Appl.
Phys. Lett. 2013, 102, 161904.
Adv. Eng. Mater. 2018, 1701159 1701159 (22 of 23)
www.aem-journal.com
[169] X. Chen, X. Zheng, J.-K. Kim, X. Li, D.-W. Lee, J. Vac. Sci. Technol. B
2011, 29, 5.
[170] M. Gamil, O. Tabata, K. Nakamura, A. M. R. F. El-Bab, A. A. El-
Moneim, Key Eng. Mater. 2014, 605, 207.
[171] C. Bonavolontà, C. Camerlingo, G. Carotenuto, S. De Nicola,
A. Longo, C. Meola, S. Boccardi, M. Palomba, G. P. Pepe,
M. Valentino, Sens. Actuators A 2016, 252, 26.
[172] J.-W. Zha, B. Zhang, R. K. Y. Li, Z.-M. Dang, Compos. Sci. Technol.
2016, 123, 32.
[173] R. Moriche, M. Sánchez, A. Jim
enez-Suárez, S. G. Prolongo,
A. Ure~ na, Compos. Sci. Technol. 2016, 123, 65.
[174] Z. Chen, W. Ren, L. Gao, B. Liu, S. Pei, H. M. Cheng, Nat. Mater.
2011, 10, 424.
[175] C. S. Boland, U. Khan, G. Ryan, S. Barwich, R. Charifou, A. Harvey,
C. Backes, Z. Li, M. S. Ferreira, M. E. Möbius, R. J. Young,
J. N. Coleman, Science 2016, 354, 1257.
[176] T. T. Tung, C. Robert, M. Castro, J. F. Feller, T. Y. Kim, K. S. Suh,
Carbon 2016, 108, 450.
[177] H. Liu, M. Dong, W. Huang, J. Gao, K. Dai, J. Guo, G. Zheng, C. Liu,
C. Shen, Z. Guo, J. Mater. Chem. C 2017, 5, 73.
[178] B. Wang, B. K. Lee, M. J. Kwak, D. W. Lee, Rev. Sci. Instrum. 2013, 84,
105005.
[179] Y.-J. Kim, J. Y. Cha, H. Ham, H. Huh, D.-S. So, I. Kang, Curr. Appl.
Phys. 2011, 11, S350.
[180] A. Tamburrano, F. Sarasini, G. De Bellis, A. G. D’Aloia, M. S. Sarto,
Nanotechnology 2013, 24, 465702.
[181] C. Acquarelli, L. Paliotta, A. Tamburrano, G. De Bellis, M. S. Sarto,
Sensors 2016, 16, 1780.
[182] Y. S. Choi, M. J. Gwak, D. W. Lee, Rev. Sci. Instrum. 2016, 87, 105004.
[183] T. T. Tung, R. Karunagaran, D. N. H. Tran, B. Gao, S. Nag-
Chowdhury, I. Pillin, M. Castro, J.-F. Feller, D. Losic, J. Mater. Chem.
C 2016, 4, 3422.
[184] V. Eswaraiah, K. Balasubramaniam, S. Ramaprabhu, Nanoscale
2012, 4, 1258.
[185] P. Costa, J. Nunes-Pereira, J. Oliveira, J. Silva, J. Agostinho Moreira,
S. A. C. Carabineiro, J. G. Buijnsters, S. Lanceros-Mendez, Comp.
Sci. Technol. 2017, 153, 241.
[186] Y. Qin, Q. Peng, Y. Ding, Z. Lin, C. Wang, Y. Li, F. Xu, J. Li, Y. Yuan,
X. He, Y. Li, ACS Nano 2015, 9, 8933.
[187] F. Nasirpouri, H. Pourmahmoudi, F. Abbasi, S. Littlejohn,
A. S. Chauhan, A. Nogaret, J. Electron. Mater. 2015, 44, 3512.
[188] C. Mattmann, F. Clemens, G. Troster, Sensors 2008, 8, 3719.
[189] C. Yan, J. Wang, W. Kang, M. Cui, X. Wang, C. Y. Foo, K. J. Chee,
P. S. Lee, Adv. Mater. 2014, 26, 2022.
[190] Y. Wang, T. Yang, J. Lao, R. Zhang, Y. Zhang, M. Zhu, X. Li,
X. Zang, K. Wang, W. Yu, H. Jin, L. Wang, H. Zhu, Nano Res. 2015,
8, 1627.
[191] L. M. Chiacchiarelli, M. Rallini, M. Monti, D. Puglia, J. M. Kenny,
L. Torre, Compos. Sci. Technol. 2013, 80, 73.
[192] R. Moriche, M. Sánchez, S. G. Prolongo, A. Jim
enez-Suárez,
A. Ure~ na, Compos. Struct. 2016, 136, 101.
[193] S. Tadakaluru, W. Thongsuwan, P. Singjai, Sensors 2014, 14, 868.
[194] F. Avil
es, A. May-Pat, M. A. L
opez-Manchado, R. Verdejo,
A. Bachmatiuk, M. H. Rümmeli, Eur. Polym. J. 2018, 99, 394.
[195] S.-H. Hwang, H. W. Park, Y.-B. Park, M.-K. Um, J.-H. Byun, S. Kwon,
Compos. Sci. Technol. 2013, 89, 1.
[196] W. Li, A. Dichiara, J. Bai, Compos. Sci. Technol. 2013, 74, 221.
[197] D. A. Walters, M. J. Casavant, X. C. Qin, C. B. Huffman, P. J. Boul,
L. M. Ericson, E. H. Haroz, M. J. O’Connell, K. Smith, D. T. Colbert,
R. E. Smalley, Chem. Phys. Lett. 2001, 338, 14.
[198] P. S. Goh, A. F. Ismail, B. C. Ng, Composites Part A 2014, 56, 103.
[199] H. Xiao, M. Liu, J. Jiang, Sens. Rev. 2015, 35, 43.
[200] H. Høyer, M. Knaapila, J. Kjelstrup-Hansen, G. Helgesen, J. Appl.
Phys. 2012, 112, 094324.
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
[201] M. Knite, A. Linarts, K. Ozols, V. Tupureina, I. Stalte, L. Lapcinskis,
Colloids Surf. A Physicochem. Eng. Asp. 2017, 526, 8.
[202] B. L. Wardle, D. S. Saito, E. J. García, A. J. Hart, R. Guzmán de
Villoria, E. A. Verploegen, Adv. Mater. 2008, 20, 2707.
[203] D. Wang, P. Song, C. Liu, W. Wu, S. Fan, Nanotechnology 2008, 19,
075609.
[204] H. Cebeci, R. Guzman de Villoria, A. J. Hart, B. L. Wardle, Compos.
Sci. Technol. 2009, 69, 2649.
[205] X. Wang, P. D. Bradford, W. Liu, H. Zhao, Y. Inoue, J.-P. Maria, Q. Li,
F.-G. Yuan, Y. Zhu, Compos. Sci. Technol. 2011, 71, 1677.
[206] Y. Inoue, Y. Suzuki, Y. Minami, J. Muramatsu, Y. Shimamura,
K. Suzuki, A. Ghemes, M. Okada, S. Sakakibara, H. Mimura,
K. Naito, Carbon 2011, 49, 2437.
[207] T. Ogasawara, S.-Y. Moon, Y. Inoue, Y. Shimamura, Compos. Sci.
Technol. 2011, 71, 1826.
[208] X. Wang, Q. Jiang, W. Xu, W. Cai, Y. Inoue, Y. Zhu, Carbon 2013, 53,
145.
[209] T. Tsuda, T. Ogasawara, S.-Y. Moon, K. Nakamoto, N. Takeda,
Y. Shimamura, Y. Inoue, Composites Part A 2014, 65, 1.
[210] S. Wu, R. B. Ladani, J. Zhang, E. Bafekrpour, K. Ghorbani,
A. P. Mouritz, A. J. Kinloch, C. H. Wang, Carbon 2015, 94, 607.
[211] L. Liu, W. Ma, Z. Zhang, Small 2011, 7, 1504.
[212] H. Garmestani, M. Al-Haik, K. Dahmen, R. Tannenbaum, D. Li,
S. S. Sablin, M. Y. Hussaini, Adv. Mater. 2003, 22, 1918.
[213] E. Camponeschi, R. Vance, M. Al-Haik, H. Garmestani,
R. Tannenbaum, Carbon 2007, 45, 2037.
[214] C. A. Martin, J. K. W. Sandler, A. H. Windle, M. K. Schwarz,
W. Bauhofer, K. Schulte, M. S. P. Shaffer, Polymer 2005, 46, 877.
[215] C. Park, J. Wilkinson, S. Banda, Z. Ounaies, K. Wise, G. Sauti,
J. Polym. Sci. B 2006, 44, 1751.
[216] C. Ma, W. Zhang, Y. Zhu, L. Ji, R. Zhang, N. Koratkar, J. Liang,
Carbon 2008, 46, 706.
Adv. Eng. Mater. 2018, 1701159 1701159 (23 of 23)
www.aem-journal.com
[217] A. I. Oliva-Avil
es, F. Avil
es, V. Sosa, Carbon 2011, 49, 2898.
[218] A. I. Oliva-Avil
es, F. Avil
es, V. Sosa, G. D. Seidel, Carbon 2014, 69,
342.
[219] A. I. Oliva-Avil
es, V. V. Zozulya, F. Gamboa, F. Avil
es, Phys. E 2016,
83, 7.
[220] E. C. Sengezer, G. D. Seidel, R. J. Bodnar, Smart Mater. Struct. 2017,
26, 095027.
[221] K. Parmar, M. Mahmoodi, C. Park, S. S. Park, Smart Mater. Struct.
2013, 22, 075006.
[222] W. Obitayo, T. Liu, Carbon 2015, 85, 372.
[223] Y. Miao, Q. Q. Yang, R. Sammynaiken, W. J. Zhang, J. Maley,
G. Schatte, Appl. Phys. Lett. 2013, 102, 233106.
[224] S. Li, J. G. Park, S. Wang, R. Liang, C. Zhang, B. Wang, Carbon 2014,
73, 303.
[225] A. Li, A. E. Bogdanovich, P. D. Bradford, Smart. Mater. Struct 2015,
24, 095004.
[226] D.-Y. Khang, J. Xiao, C. Kocabas, S. MacLaren, T. Banks, H. Jiang,
Y. Y. Huang, J. A. Rogers, Nano Lett. 2008, 8, 124.
[227] M. A. Cullinan, M. L. Culpepper, Phys. Rev. B 2010, 82, 115428.
[228] A. Lekawa-Raus, K. K. K. Koziol, A. H. Windle, ACS Nano 2014, 8,
11214.
[229] K. Aly, A. Li, P. D. Bradford, Composites Part A 2016, 90, 536.
[230] J. R. Gaier, M. E. Slabe, Carbon 1990, 28, 669.
[231] Q. Lu, G. Keskar, R. Ciocan, R. Rao, R. B. Mathur, A. M. Rao,
L. L. Larcom, J. Phys. Chem. B 2006, 110, 24371.
[232] M. D. Rein, O. Breuer, H. D. Wagner, Compos. Sci. Technol. 2011, 71,
373.
[233] S. Luo, W. Obitayo, T. Liu, Carbon 2014, 76, 321.
[234] P. Slobodian, P. Riha, R. Olejnik, U. Cvelbar, P. Saha, Compos. Sci.
Technol. 2013, 81, 54.
[235] A. Nakamura, T. Hamanishi, S. Kawakami, M. Takeda, Mater. Sci.
Eng. B 2017, 219, 20.
© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim