Diamond & Related Materials 65 (2016) 3236

Contents lists available at ScienceDirect

Diamond & Related Materials

journal homepage: www.elsevier.com/locate/diamond

Low-frequency electric noise spectroscopy in different polymer/carbon

nanotubes composites
C. Barone a,b, G. Landi c, C. Mauro a,b,, S. Pagano a,b, H.C. Neitzert c
a
b
c

Dipartimento di Fisica E.R. Caianiello, Universit di Salerno, 84084 Fisciano, Salerno, Italy
CNR-SPIN Salerno, Universit di Salerno, 84084 Fisciano, Salerno, Italy
Dipartimento di Ingegneria Industriale, Universit di Salerno, 84084 Fisciano, Salerno, Italy

a r t i c l e

i n f o

Article history:
Received 29 October 2015
Received in revised form 11 December 2015
Accepted 5 January 2016
Available online 7 January 2016
Keywords:
Polymer/carbon nanotubes composites
Electrical noise characterization
Sensors
Percolation

a b s t r a c t
Carbon nanotubes addition to polymer and epoxy material allows to realize a large variety of new types of sensors and actuators. For the development of these devices, a deeper understanding of the basic charge carriers
transport mechanisms is needed and low-frequency noise spectroscopy can effectively contribute to this task.
The dc and electrical noise characteristics of different polymer/carbon nanotubes composites are analyzed at
temperatures from 10 to 300 K. It has been found that a random tunnel junctions resistive networks model
describes all the systems under test. A crossover from a two-level tunneling mechanism, at low temperatures,
to resistance uctuations induced by percolative paths, in the high-temperature region, has also been observed.
This behavior of the 1/f noise is a general feature of the investigated highly conducting nanocomposites, independent of the specic polymer matrix and nanotube concentration.
2016 Elsevier B.V. All rights reserved.

1. Introduction
Only few years after their discovery [1], carbon nanotubes (CNTs)
have been used in a wide range of electrical applications. For the realization of CNT based transistors, single-wall carbon nanotubes (SWCNTs)
are preferred [2], and recently thin lm transistors (TFTs) made from
parallel arrays [3] or networks [4] of semiconductor-enriched SWCNTs
have reached performances, that not only rival, but outperform the
commercial state-of-the-art thin lm transistor technologies.
Multi-wall carbon nanotubes (MWCNTs), on the other hand, generally
have metallic character that play an important role in other applications
with and without a polymeric matrix, as transparent conductive
contacts for solar cells [5], for RF-shielding [6] and as sensors and
actuators.
Electrical noise is generally considered as a limiting factor for the
functionality of electronic devices and sensors. Therefore, the understanding of the underlying mechanisms, useful to lower the noise
level, is of great interest. Moreover, noise can also be an additional critical criterion in sensing applications. This technique, called uctuation
induced sensing [7], has also been applied to carbon nanotube structures [8]. 1/f-type noise has been reported in SWCNTs based ballistic
transistors and attributed to the intrinsic current uctuations, due to
the inuence of nearby charge-traps [9]. Also for SWCNT network
based transistors, the 1/f-type noise has been observed and has been
related to the mobility change due to adsorption and desorption of gas
Corresponding author.
E-mail address: cmauro@unisa.it (C. Mauro).

http://dx.doi.org/10.1016/j.diamond.2016.01.004
0925-9635/ 2016 Elsevier B.V. All rights reserved.

molecules to the CNT surface [10]. While there are a series of studies regarding the noise behavior of devices prepared with SWCNTs [911],
fewer reports on noise characteristics of MWCNT composites can be
found. In this respect, it has been reported that the noise spectrum for
metallic CNT based gated structures has also 1/f behavior, and its amplitude is up to 2 orders of magnitude lower than the one measured in
semiconducting SWCNTs [12].
Supported by results obtained through the noise characterization of
innovative and complex systems such as manganites [13], iron chalcogenides and pnictides superconductors [14,15], and organic photovoltaic devices [16,17], voltage-spectral density measurements have been
carried out on different polymer/MWCNT composites. The dominant
contributions to the overall electrical resistance are reported in this
paper. In addition, the low-frequency noise behavior of a highly conductive epoxy/MWCNT compound has been compared to that of samples
with High-density polyethylene (HDPE)/MWCNT and Low-density
polythylene (LDPE)/MWCNT composites, in order to check the existence of similar transport mechanisms. This study could be of interest
for these types of materials, very attracting for sensing and actuating
applications. For example, the epoxy/MWCNT composite has already revealed its potential as a very stable heating element with temperature
self-sensing functionality [18].
2. Experimental
The composites here investigated were prepared by mixing CNTs in
different polymeric matrices. MWCNTs were produced by chemical
vapor deposition at CSIRO (Commonwealth Scientic and Industrial

C. Barone et al. / Diamond & Related Materials 65 (2016) 3236

Research Organization, Australia) and dispersed in two polyethylene

bases, with an average molecular weight of ~60000 g/mol and a polydispersity of 5.8 (High-density polyethylene HDPE0390 from Qenos)
and ~115000 g/mol and a polydispersity of 7.6 (Low-density polyethylene LDPE from Scientic Polymer Products), respectively, with two
MWCNT concentrations (5 and 7 wt%). Synthesized MWCNTs by catalytic carbon vapor deposition process (Nanocyl S.A.) were incorporated
in a diglycidil-ether bisphenol-A (DGEBA) epoxy resin, with a concentration of 0.5 wt%. Polymeric matrix compound is usually considered
as insulating material due to its high electrical resistivity (1010 - 1015
m). Adding conductive material into the insulating matrix can lead to
an overall conductive compound, depending on the volume fraction of
the conductive phase. In particular, above a threshold volume fraction,
the resistivity of the composite dramatically drops by several orders of
magnitude due to percolation. To sum up, CNTs give several features
to the composites: even at low loading, CNTs act as a conductive additive, thanks to their low percolation thresholds (between 1 and
2.5 wt% for HDPE and LDPE, and lower than 0.1 wt% for epoxy), while
in larger percentage enhance mechanical properties of the polymer, if
dispersed uniformly [19]. The detailed information reported in [20]
and [21] indicate that all the used polymer/carbon nanotubes compounds are suitable for sensor and conductor applications, due to their
high conductivity values. This behavior of the composites seems to be
not inuenced by the presence of a bundle organization which, however, has been observed through a scanning electron microscopy analysis.
The experimental setup provides both two- and four-probe connections to the samples through evaporated gold contact pads (25-nmthick, 5-mm-wide and about 1-mm-distant). Measurements were
carried out in a closed-cycle refrigerator, with a temperature range of
operation from 10 to 325 K. The temperature was stabilized by a resistance heater controlled in a closed feedback loop, giving a stability
better than 0.1 K. A low-noise Keithley dc current source biased the
samples. The dc voltage drop was recorded with a digital multimeter,
while the ac voltage signal was amplied with a low-noise PAR5113
preamplier and analyzed by a dynamic signal analyzer HP35670A. Possible effects of spurious external noise sources were eliminated by
resorting to appropriate techniques, accurately described in [22].
3. Results and discussion
3.1. Electric transport measurements
The temperature dependence of the resistance R for the ve investigated MWCNT composites is shown in Fig. 1. The evident monotonic

33

decrease with increasing temperature can be explained by using different electrical transport mechanisms, such as: Luttinger liquid [23]; variable range hopping [24]; and uctuation-induced tunneling, proposed
by Sheng [25]. All these models are able to reproduce the experimental
data. However, an accurate statistical analysis indicates that the lowest
reduced 2 and the highest coefcient of determination values are given
by the Sheng model. In this framework, frequently considered in the
case of carbon nanotubes compounds [26,27], a uctuation-induced
tunneling conduction between conducting regions (the MWCNTs) and
small insulating barriers (the matrix) is modeled through the following
expression



T1
;
RT R0 exp
T T0

where R0 is a proportionality coefcient, and T0 and T1 are two characteristic temperatures of the systems under test [25]. The best t
curves by using Eq. (1) are shown in Fig. 1 as dashed lines, revealing a
good agreement with the data points in the whole temperature range
and for all the samples. The estimation of T1, reported in Table 1 together with all the other tting parameters, allows the evaluation of the
minimum energy (b kBT1) necessary to cross the insulating barriers
formed among the nanotube aggregates [26]. These values in Table 1
are similar to those observed for other carbonaceous materials, analyzed in terms of the Sheng model [27].
A parallel conduction mechanism of thermal activation over the potential barrier can not be neglected in the high-temperature region. This
process is described by


TcT1
1  A 2 ;
RT R1 exp
T T c T 0

where R1 is a preexponentional factor, A is the dimensionless applied electric eld, and Tc is a crossover temperature above which the
uctuation-induced tunneling and thermal activation contributions
are indistinguishable [25]. The curves with Eq. (2), shown in Fig. 1 as
solid lines, are obtained with the values of Tc reported in Table 1.
These crossover temperatures, associated to the energy distribution of
the tunnel barriers, indicate that for the same polymeric matrices the
higher values are related with the lowest MWCNT concentrations. This
is usually expected, when a modication in the interaction between
the polymeric matrix and the carbonaceous ller occurs. Such a conclusion is not invalidated by the presence of a small correction factor A in
Eq. (2), which gives, as a rst approximation, a negligible contribution
to the evaluation performed. However, the dc analysis, alone, does not
identify the transport mechanism of charge carriers in the hightemperature regime. Therefore, a more sensitive investigation is
needed.
3.2. Electric noise measurements

Fig. 1. Temperature dependence of the resistance R for the ve different types of

investigated MWCNT composites. The best tting curves to the experimental data are
also reported as dashed lines [uctuation-induced tunneling, see Eq. (1)] and solid lines
[thermal activation mechanism, see Eq. (2)]. Above the crossover temperature Tc, the
two theoretical models are indistinguishable.

When electric noise is involved, the spectral density of voltage

uctuations SV must be examined. The frequency dependence of SV
is visible in Fig. 2(a) for the ve different types of MWCNT compounds
at the reference temperature of 300 K. All the spectra can be regarded
as superposition of two main components: the rst determines a 1/f
dependence in the low-frequency region and the second, characterized
by a constant spectrum at higher frequencies, is due to the
Johnson noise 4kBTR added to the background electronic chain noise
(~1 10-17 V 2/Hz).
In order to extract information on the dynamics of uctuations and
the conduction mechanisms, it is useful to study the 1/f contribution
versus the applied bias current. This dependence is shown in Fig. 2(b),

34

C. Barone et al. / Diamond & Related Materials 65 (2016) 3236

Table 1
Relevant parameters extracted from the dc measurements of the investigated MWCNT composites: proportionality factors R0 and R1; characteristic temperatures T0 and T1; energetic levels
b kBT1; crossover temperatures Tc; dimensionless applied electric eld A. The reported error values are obtained by using a standard statistical analysis associated to the tting
procedure.
composite

R0 (k)

R1 (k)

T0 (K)

T1 (K)

b (meV)

Tc (K)

LDPE +5 wt%
LDPE +7 wt%
HDPE +5 wt%
HDPE +7 wt%
epoxy +0.5 wt%

31.20.4
20.10.4
19.50.4
1.310.03
0.580.01

51.40.9
36.30.8
19.80.5
1.290.03
0.640.01

1061
981
65.00.8
32.20.5
59.10.7

2784
2904
1572
971
861

23.90.3
24.90.3
13.50.2
8.30.1
7.40.1

1374
1093
N300
1775
1003

0.3340.007
0.3510.007
0.2810.006
0.3040.006
0.2630.005

revealing a quadratic behavior for all the samples under test and, as a
consequence, resistance uctuations as the dominant noise source. In
this context, the noise level of Ohmic compounds can be expressed by
using the Hooge relationship as [28]
Noise Level

f SV
2 2

R I

H 1
;
n s

where H is the dimensionless Hooge material-dependent constant,

n the carrier density, and s the sample volume.
3.3. Theoretical discussion
As already reported in literature [2931], the best way to model the
noise level in this kind of disordered systems is given by

Noise Level

8
>
A T R2
>
<

TbT c ;

>
S Rw
>
: 0

TNT c ;

where A is a proportionality factor, S0 is the amplitude of resistance

uctuations, w takes into account the critical indexes for lattice percolation, and Tc is the crossover temperature between the uctuationinduced tunneling and thermal activation regimes. The parameters A
and S0, alone, do not give the intrinsic noise of the material, which is
further dependent on the sample temperature, resistance, and volume.
In details, below Tc, two-level tunneling system (TLTS) uctuations
dominate, while above Tc thermal activation must be considered, as a
consequence of the percolation model [31]. It is important to stress
here, that the two-level tunneling mechanism has successfully been
used for the interpretation of the noise in individual nanotube samples
[32], and also for HDPE compounds below the crossover temperature
[30]. The experimental noise level, compared with the theoretical
formulation given by Eq. (4), is shown in Fig. 3(a). Below Tc, the good
agreement of TLTS model (dashed lines) with the data points is evident.
Simultaneously, the percolative uctuation process (solid lines) is

dominant in the temperature region above Tc. Such a behavior is independent of the polymer matrix chosen for the device realization.
By comparing Eq. (4) with Eq. (3), it is possible to extract the
normalized Hooge parameter as
8
2
>
> A s T R
H <

>
n
S Rw
>
: 0 s

TbT c ;
TNT c :

The exponent w assumes values of 1 for epoxy, 1.01 for LDPE, and
1.06 for HDPE, which are very close to the values predicted by the classical percolation theory in the case of three-dimensional systems [33].
By considering s as the volume of the entire composite, that is the
polymeric matrix and the conductive nanotubes, H/n is estimated
with Eq. (5) and the results are shown in Fig. 3(b). The alternative
choice of assuming s as the volume of the conducting region, alone,
is equally valid. However, by taking into account the used MWCNT concentrations, the rescaling of the data in Fig. 3(b) does not signicantly
change the reported experimental behavior. In this respect, it is worth
noting that, corresponding to Tc, the normalized Hooge parameter
exhibits a maximum, probably related to the change of the charge
carrier transport properties. The main conclusion of this analysis is
that, the epoxy based composite seems to be the most promising for
sensor applications, due to its lowest noise level in the whole investigated temperature range.
4. Conclusions
Highly conductive HDPE, LDPE, and epoxy resin loaded with different
MWCNT concentrations have been investigated. Both the temperaturedependent electrical conductance and the voltage-noise measurements
indicate that the electrical resistance of the compounds is dominated, in
the low-temperature regime, by the charge carrier tunneling between
adjacent carbon nanotubes. This behavior is independent of the specic
polymeric matrix. On the other hand, above a crossover temperature,
specic for each composite, a thermally activated transport mechanism

Fig. 2. Spectral traces of all the investigated MWCNT composites at a reference temperature of 300 K and bias current of 4 A (a). The quadratic bias dependence of the experimental
voltage-spectral density at 90 Hz is also shown at the same reference temperature of 300 K (b).

C. Barone et al. / Diamond & Related Materials 65 (2016) 3236

35

Fig. 3. (a) Temperature dependence of the Noise Level for the ve investigated MWCNT composites. The best tting curves obtained by using a two-level tunneling system model [see
Eq. (4), T b Tc] and a percolative noise mechanism [see Eq. (4), T N Tc] are shown as dashed and solid lines, respectively. The transition between these two regimes occurs at the
crossover temperature, estimated from the dc analysis. In the case of HDPE +5 wt% MWCNT composite, the percolative model is not shown, being out from the temperature range
considered (the value of Tc is higher than 300 K). (b) Temperature dependence of the normalized Hooge parameters computed with Eq. (5).

is dominant. In particular, noise spectroscopy clearly reveals the transition between these two regimes. An universal crossover from a TLTS uctuation process at low temperatures to resistance uctuations in a
percolative network at higher temperatures is observed. The lowest
noise level in the whole investigated temperature range has been found
for the epoxy resin based composites. This adds to the good high temperature stability and to the low percolation threshold of this latter material,
making it very promising for demanding applications.
Prime novelty statement
Prime novelty of the work lies in the following points:
- Voltage-spectral density and electric noise properties are reported
for carbon nanotubes composites.
- The noise spectroscopy gives a better understanding of the charge
carrier transport in carbon nanotubes compounds.
- The analysis of the uctuation mechanisms reveals that the
crossover between two different transport regimes is a general and
universal feature of the investigated highly conducting polymer/
nanotubes composites, not depending on the polymer matrix and
on the nanotube concentrations.

Acknowledgements
The authors would like to thank S. Abate of CNR-SPIN Salerno for his
valuable technical support. O. Valentino, M. R. Nobile (from Salerno University), and G. Simon (from Monash University) are acknowledged for
the preparation and structural characterization of the polyethylene
based composites. L. Guadagno, L. Vertuccio, and A. Sorrentino (from Salerno University) are acknowledged for the preparation and structural
characterization of the epoxy based composites.
References
[1] S. Iijima, Helical microtubules of graphitic carbon, Nature 354 (1991) 5658.
[2] P. Avouris, R. Martel, V. Derycke, J. Appenzeller, Carbon nanotube transistors and
logic circuits, Physica B 323 (2002) 614.
[3] G.J. Brady, Y. Joo, S.S. Roy, P. Gopalan, M.S. Arnold, High performance transistors via
aligned polyuorene-sorted carbon nanotubes transistors, Appl. Phys. Lett. 104
(2014) 083107.
[4] W. Xue, T. Cui, Thin-lm transistors with controllable mobilities based on layer-bylayer self-assembled carbon nanotube composites, Solid State Electron. 53 (2009)
10501055.
[5] J.S. Moon, J.H. Park, T.Y. Lee, Y.W. Kim, J.B. Yoo, C.Y. Park, J.M. Kim, K.W. Jin, Transparent conductive lm based on carbon nanotubes and PEDOT composites, Diam. Relat.
Mater. 14 (2005) 18821887.
[6] M.H. Al-Saleh, Inuence of conductive network structure on the EMI shielding and
electrical percolation of carbon nanotube/polymer nanocomposites, Synth. Met.
205 (2015) 7884.

[7] L.B. Kish, R. Vajtai, C.G. Granqvist, Extracting information from noise spectra of
chemical sensors: single sensor electronic noses and tongues, Sensors Actuators B
Chem. 71 (2000) 5559.
[8] I. Lee, X. Liu, C. Zhou, B. Kosko, Noise-enhanced detection of subthreshold signals
with carbon nanotubes, IEEE Trans. Nanotechnol. 5 (2006) 613627.
[9] J. Tersoff, Low-frequency noise in nanoscale ballistic transistors, Nano Lett. 7 (2007)
194198.
[10] U.J. Kim, K.H. Kim, K.T. Kim, Y.-S. Min, W. Park, Noise characteristics of single-walled
carbon nanotube network transistors, Nanotechnology 19 (2008) 285705.
[11] M. Briman, K. Bradley, G. Gruner, Source of 1/f noise in carbon nanotube devices, J.
Appl. Phys. 100 (2006) 013505.
[12] S. Reza, Q.T. Huynh, G. Bosman, J. Sippel-Oakley, A.G. Rinzler, 1/f noise in metallic
and semiconducting carbon nanotubes, J. Appl. Phys. 100 (2006) 094318.
[13] P. Orgiani, C. Adamo, C. Barone, A. Galdi, S. Pagano, A.Yu. Petrov, O. Quaranta, C.
Aruta, R. Ciancio, M. Polichetti, D. Zola, L. Maritato, Epitaxial growth of
La0.7Ba0.3MnO3 thin lms on MgO substrates: structural, magnetic, and transport
properties, J. Appl. Phys. 103 (2008) 093902.
[14] C. Barone, S. Pagano, I. Pallecchi, E. Bellingeri, M. Putti, C. Ferdeghini, Thermal and
voltage activated excess 1/f noise in FeTe0.5Se0.5 epitaxial thin lms, Phys. Rev. B
83 (2011) 134523.
[15] C. Barone, F. Romeo, S. Pagano, M. Adamo, C. Nappi, E. Sarnelli, F. Kurth, K. Iida, Probing transport mechanisms of BaFe2As2 superconducting lms and grain boundary
junctions by noise spectroscopy, Sci. Rep. 4 (2014) 6163.
[16] G. Landi, C. Barone, A. De Sio, S. Pagano, H.C. Neitzert, Characterization of polymer:
fullerene solar cells by low-frequency noise spectroscopy, Appl. Phys. Lett. 102
(2013) 223902.
[17] C. Barone, G. Landi, A. De Sio, H.C. Neitzert, S. Pagano, Thermal ageing of bulk
heterojunction polymer solar cells investigated by electric noise analysis, Sol. Energy Mater. Sol. Cells 122 (2014) 4045.
[18] H.C. Neitzert, L. Vertuccio, A. Sorrentino, Epoxy/MWCNT composite as temperature sensor and electrical heating elements, IEEE Trans. Nanotechnol. 9 (2011)
688693.
[19] B.P. Singh, P. Prabha, T. Saini, P. Gupta, G. Garg, I. Kumar, S. Pande, R.K. Pande, S.K.
Seth, R.B.Dhawan Mathur, Designing of multiwalled carbon nanotubes reinforced
low density polyethylene nanocomposites for suppression of electromagnetic radiation, J. Nanoparticle Res. 13 (2011) 70657074.
[20] O. Valentino, M. Sarno, N.G. Rainone, M.R. Nobile, P. Ciambelli, H.C. Neitzert, G.P.
Simon, Inuence of the polymer structure and nanotube concentration on the conductivity and rheological properties of polyethylene/CNT composites, Phys. E. 40
(2008) 24402445.
[21] L. Guadagno, L. Vertuccio, A. Sorrentino, M. Raimondo, C. Naddeo, V. Vittoria, G.
Iannuzzo, E. Calvi, S. Russo, Mechanical and barrier properties of epoxy resin lled
with multi-walled carbon nanotubes, Carbon 47 (2009) 2419.
[22] C. Barone, A. Galdi, S. Pagano, O. Quaranta, L. Mchin, J.-M. Routoure, P. Perna, Experimental technique for reducing contact and background noise in voltage spectral
density measurements, Rev. Sci. Instrum. 78 (2007) 093905.
[23] M. Bockrath, D.H. Cobden, J. Lu, A.G. Rinzler, R.E. Smalley, L. Balents, P.L. McEuen,
Luttinger-liquid behaviour in carbon nanotubes, Nature 397 (1999) 598601.
[24] N.F. Mott, E.A. Davis, Electronic Processes in Non-Crystalline Materials, Clarendon,
Oxford, 1979.
[25] P. Sheng, Fluctuation-induced tunneling conduction in disordered materials, Phys.
Rev. B 21 (1980) 21802195.
[26] A. Mdarhri, F. Carmona, C. Brosseau, P. Delhaes, Direct current electrical and microwave properties of polymer-multiwalled carbon nanotubes composites, J. Appl.
Phys. 103 (2008) 054303.
[27] A. Mdarhri, C. Brosseau, A review and analysis of microwave absorption in
polymer composites lled with carbonaceous particles, J. Appl. Phys. 111
(2012) 061301.
[28] S. Kogan, Electronic Noise and Fluctuations in Solids, Cambridge University Press,
Cambridge, 1996.
[29] C. Barone, S. Pagano, H.C. Neitzert, Effect of concentration on low-frequency noise of
multiwall carbon nanotubes in high-density polyethylene matrix, Appl. Phys. Lett.
97 (2010) 152107.

36

C. Barone et al. / Diamond & Related Materials 65 (2016) 3236

[30] C. Barone, S. Pagano, H.C. Neitzert, Transport and noise spectroscopy of MWCNT/
HDPE composites with different nanotube concentrations, J. Appl. Phys. 110
(2011) 113716.
[31] C. Barone, G. Landi, C. Mauro, H.C. Neitzert, S. Pagano, Universal crossover of the
charge carrier uctuation mechanism in different polymer/carbon nanotubes
composites, Appl. Phys. Lett. 107 (2015) 143106.

[32] D. Tobias, M. Ishigami, A. Tselev, P. Barbara, E.D. Williams, C.J. Lobb, M.S. Fuhrer, Origins of 1/f noise in individual semiconducting carbon nanotube eld-effect transistors, Phys. Rev. B 77 (2008) 033407.
[33] V. Podzorov, M. Uehara, M.E. Gershenson, T.Y. Koo, S.-W. Cheong, Giant 1/f noise in
perovskite manganites: evidence of the percolation threshold, Phys. Rev. B 61
(2000) R3784R3787.