Composites Science and Technology 104 (2014) 104–110

Contents lists available at ScienceDirect

Composites Science and Technology

journal homepage: www.elsevier .com/ locate /compsci tech

Skin effect of conductive polymer composites observedby high-resolution solid-state NMR

http://dx.doi.org/10.1016/j.compscitech.2014.08.0310266-3538/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Université de Lyon, Université Lyon 1, IMP@LYON1,UMR CNRS 5223, 15 Boulevard A Latarjet, Villeurbanne Cedex 69622, France.

E-mail address: jouni.mohammad@gmail.com (M. Jouni).

Mohammad Jouni a,b,⇑, Anton Buzlukov c, Michel Bardet d, Fernande Da Cruz-Boisson b,Asma Eddarir d, Valérie Massardier b, Gisèle Boiteux a

a Université de Lyon, Université Lyon 1, IMP@LYON1, UMR CNRS 5223, 15 Boulevard A Latarjet, Villeurbanne Cedex 69622, Franceb INSA de Lyon, IMP@INSA, UMR CNRS 5223, 20 Avenue Albert Einstein, Bat J. Verne, Villeurbanne Cedex 69621, Francec Laboratory of Kinetic Phenomena IMP UB RAS, 620990 Ekaterinburg, Russiad Univ. Grenoble Alpes, INAC-SCIB, LRM, F-38000 Grenoble, France, CEA, INAC-SCIB, LRM, F-38000 Grenoble, France

a r t i c l e i n f o

Article history:Received 16 July 2014Received in revised form 27 August 2014Accepted 31 August 2014Available online 6 September 2014

Keywords:A. Nano particlesA. Nano compositesB. Electrical propertiesD. Solid state NMRE. Extrusion

a b s t r a c t

High-resolution solid-state Nuclear Magnetic Resonance (NMR) combined with other investigations wasapplied to provide essential information on conductive polymer composites based on high-density poly-ethylene (HDPE) as matrix and multi-walled carbon nanotubes (MWCNTs) or silver nanoparticles (Ag-NPs) as fillers. All composites were prepared by melt mixing using an extrusion process and characterizedelectrically. By studying the general features of NMR spectra and the molecular dynamic from NMR relax-ation parameters, it was possible to obtain structural information about the organization and dispersionof fillers. Due to the paramagnetic or conductive nature of the fillers, it was found that a loss of NMR sig-nal occurred with increasing amounts of filler. In the case of Ag-NPs, this phenomenon was attributed to askin effect caused by the conductive properties of fillers limiting the adsorption of radiofrequency.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Conductive polymer composites (CPCs) have been the object ofintense researches recently [1–5]. The large potential applicationsand advantages of these composites mean they are privileged overmany other materials [6–9]. Indeed, CPCs can serve in many sec-tors involving particular physical characteristics such as electricdissipation and shielding of electromagnetic interference (EMI)for use in automotive and aerospace domains.

Despite the large number of papers published in this area [10–14], there are still some important investigations which have notbeen published, as most studies have reported the electrical prop-erties of these composites as basic results. A better description canbe given when the later are interpreted in the scope of polymermicrostructure and the organization/dispersion of conductive fill-ers. Some papers discuss the application of NMR as a characteriza-tion technique for interactions between filler and matrix inpolymer composites [15–19], but there are very few reports con-cerning CPCs [20]. A reason for this limitation could be due tothe conductive filler itself, as reported in a previous study of poly-

urethane/carbon-fiber composites where the paramagnetic resi-dues of carbon fibers caused severe difficulties for NMRobservations [20]. Also, there are no papers focusing on the charac-terization of CPCs when metallic fillers are involved. We believethat solid state NMR can also be applied in this case to completeinvestigations carried out with other techniques (electrical mea-surements, morphology, and microstructure. . .), giving informa-tion at the molecular level on the interactions between polymerchains and fillers when studying the chemical shifts and the relax-ation parameters of the polymer matrix [21–23].

In this work, 13C NMR spectra were recorded using either directexcitation (HPDEC-MAS) or proton-to-carbon cross-polarization(CP-MAS) experiments under magic angle spinning on two typesof CPCs (PE/MWCNT and PE/Ag-NPs). The effect of the incorpora-tion of the conductive fillers on the chemical shift and line widthwas discussed. From the 13C magnetization build-up curvesrecorded under CP MAS condition as a function of the CP contacttime, the proton spin lattice relaxation time in the rotating frame(T1qH), and the cross-polarization transfer time (TCH) were com-puted. These time constants are known to be related to the molec-ular dynamics of the polymer chains [24]. Moreover, the effect ofthe amounts of conductive fillers on those constants and on thewhole observed NMR signal was also studied in detail.

Table 1Composition of studied composites.

Series PE/MWCNT PE/Ag-NPS

Filler loading (vol.%) 0 0.5 8.5 0 2 5 10 20

M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110 105

2. Experimental

2.1. Sample preparation and composition

Two series of polymer composites were investigated by solid-state NMR in this study. The first was high-density polyethylene(HDPE) filled with multi-walled carbon nanotubes (MWCNTs),which are conductive fillers with a high aspect ratio (l/d � 160),and the second series was HDPE filled with silver nanoparticles(Ag-NPs) with an average diameter of �100 nm. All compositeswere produced in melt state by extrusion using a co-rotatingtwin-screw mini-extruder (Micro 15 Twin-Screw DSM research)to obtain conductive polymer composites. More details about thepreparation and characterization can be found in [25,26].

Table 1 presents the composition of the samples studied bysolid-state NMR. The filler loading is given as a volume fraction,which is more reliable for the explanation and interpretation ofthe data obtained from NMR and electrical conductivitycharacterizations.

2.2. Experimental RMN setup

High-resolution solid-state 13C NMR spectra were mainlyrecorded on a Bruker Avance DSX 200 MHz spectrometer operatingat 50.3 MHz for 13C, using a combination of the proton-to-carboncross-polarization, high-power proton decoupling, and magic anglespinning (CP/MAS) methods. The spinning rate was set at 5000 Hz.The 1H radio-frequency field strength was set to give a protonpulse duration of around 3.103 us; the same value was used forthe dipolar decoupling process. The 13C radio frequency fieldstrength was obtained by matching the Hartman–Hahn condition.Records of 1024 transients with contact time and recycle delay of1 ms and 3 s, respectively, represented standard conditions. Thechemical shift values were obtained via the glycine carbonyl signal,which was set at 176.03 ppm relative to tetramethylsilane (TMS).In order to obtain quantitative data, 13C spectra were recordedwith direct 13C excitation with high-power proton decoupling dur-ing the NMR signal acquisition experiment. A 60 s recycling delaywas used to assure complete relaxation of 13C magnetizations.

In order to obtain T1qH and TCH values, the carbon magnetizationbuild-up under CP conditions was performed by varying the con-tact time values from 10 ls to 20 ms and worked out as describedin the experimental section of one of our previously published arti-cles [23]. From these experiments, T1qH and TCH were computed byusing the simplex fitting program (SIMFIT) provided by Bruker andOrigin software. Dilution was performed by grinding the sample ina mortar with silica. Both compounds were precisely weighed, andthe amount of material used to fill the 7 mm diameter rotors wasalso weighed. Therefore, the NMR signal areas were normalizedtaking into account the mass of polymer in each rotor and thenumber of transients acquired for each NMR experiment. Spectrumdeconvolution was carried out with the DIMFIT program developedby Massiot et al. [27].

Fig. 1. HPDEC-MAS solid state NMR spectrum of the neat high-density polyethyl-ene. Lines 1 and 2 represent the contributions of crystalline and amorphous phasesrespectively.

3. Results and discussion

3.1. General features of 13C NMR spectra

The 13C HPDEC-MAS spectrum of the pristine HDPE is shown inFig. 1. A 60 s recycling delay was used, which allows quantitativedata to be recorded. Deconvolution of the full signal leads to twodistinct peaks (green and blue lines) at 32.4 and 30.6 ppm,assigned to the crystalline and amorphous phases respectively.Such attribution was confirmed by the CPMAS spectrum (notshown here), in which the carbon signal intensity of the crystalline

part was clearly enhanced. The crystalline domain contributionwas found to be larger than the amorphous one. This is consistentwith the results from X-ray and DSC experiments, which showed ahigh crystalline content in this PE.

Fig. 2a and b shows the HPDEC-MAS spectra of two polyethylenecomposites, HDPE/MWCNT loaded with 8.5 vol.% and PE/Ag-NPsloaded with 2 vol.%. As the whole NMR signal has been normalizedfor all samples to the amount of polymer inside the rotor, by com-paring it with the spectrum of the pristine polyethylene (Fig. 1), sig-nificant decreases of the signal intensities were recorded for most ofthe filled PE (see Table 2). However, a third peak with very smallintensity was also observed around 33 ppm in the case of PE/Ag-NP composites. This peak was almost undetectable in the unfilledPE because of its very low intensity. The origin of this peak couldbe related to the monoclinic phase of HDPE as reported previously[28].

3.2. Line widths, chemical shifts, and normalized areas

The values of signal chemical shifts and widths for all compos-ites are listed in Table 2. On one hand, the chemical shifts of carbonsignals for all lines were not altered by the incorporation ofconductive fillers in both types of composites. This means thatthe chemical environments of the analyzed carbon atoms are notdrastically modified and no strong interactions between polymermatrix and fillers can be depicted. On the other hand, addition offillers leads to a broadening of the signals in both cases. Such find-ings have been reported for thermoplastic polyurethane matrixfilled with carbon fibers and assigned to the difference in magneticsusceptibility between the fillers and polymer matrix [20]. Notethat such interactions should be averaged by the spinning at themagic angle, which means that the broadening is probably due tohyperfine interactions. The broadening of signals when theMWCNT concentration increases compared to a merely stablewidening in the case of PE/Ag-NPs indicates different behaviors

Fig. 2. 13C HPDEC-MAS spectrum of (a) PE/MWCNT (8.5 vol.%) and (b) PE/Ag-NPs (2 vol.%).

Table 2Chemical shifts, widths, and normalized areas of 13C NMR signals for PE/MWCNT and PE/Ag-NP composites. Values are taken from 13C HPDEC-MAS NMRexperiments.

Sample Line 1 signal (crystalline phase) Line 2 signal (amorphous phase) Line 3 signal Normalized entire area

Shift (ppm) Width (Hz) Shift (ppm) Width (Hz) Shift (ppm) Width (Hz)

PE 32 47.5 31 103.4 NA NA 10.5-CNT 32 51.0 31 113.3 NA NA 18.5-CNT 32 109.7 31 241.7 NA NA 0.712-Ag 32 53.5 31 165.6 33 78.0 0.835-Ag 32 45.1 31 149.4 33 75.0 0.8110-Ag 32 46.9 31 179.4 33 69.0 0.7920-Ag 32 61.8 31 167.3 34 43.9 0.51

106 M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110

of fillers in both types of composites. In the case of PE/MWCNTsamples, the intrinsic paramagnetic properties of MWCNTs areclearly the origin of this broadening. However, in the case of PE/Ag-NPs, an increase in the line width is recorded after the additionof Ag-NPs, but this variation is not proportional to the filler con-tent. When adding Ag-NPs, the width of line 2 corresponding tothe amorphous phase is much broader than the one correspondingto the crystalline phase. In order to better understand the struc-tural changes caused by the addition of nanoparticles, the protonspin–lattice relaxation time in the rotating frame T1qH and thecross-polarization transfer time TCH were measured for bothcomposites.

3.3. Proton spin–lattice relaxation T1qH and cross-polarization TCH

times

The fitting of the NMR data was carried out with either aone-component or a two-component model [29,30]. The two-component model is more realistic since it takes into account thehypothesis of the occurrence of two molecular domains, whichcan be easily justified on the basis of the qualitative observationof signals assigned to the amorphous and crystalline domains.Indeed, fits are much better with the two-component model. How-ever, even with a two-component model, both domains can becharacterized with a single T1qH. That indicates that the protonsof both domains are interacting through their dipolar coupling,leading to proton spin diffusion in the entire material. This behav-ior is observed for all the samples. On the other hand, two TCH val-ues, TCHlong and TCHshort for lines 2 and 1 respectively, are necessary

to fit the experimental data recorded by NMR. All measured valuesof T1qH and TCH are reported in Table 3.

The TCH values of the visible part for the composites of PE/MWCNT do not show a significant change with the filler concen-tration and are in the same range as those of the neat polyethyl-ene. Moreover, the relative ratios of both domains remain in thesame range of the starting PE, 0.3–0.48. In the case of PE/Ag-NPcomposites, the values of TCHlong increase significantly, while adrastic decrease in the values of TCHshort can be noticed; thesechanges appear to be rather independent of the amount of filler.Moreover, the ratio of the molecular domain corresponding tolong TCH increased significantly compared with the starting mate-rial, and is found to be almost identical to 0.8 for all the concen-trations of Ag-NPs. This increase in TCHlong values indicates anincrease in molecular mobility in the amorphous part of the com-posites, despite the addition of hard metallic particles, while thecrystalline domains appear to become more rigid with shorterTCH values [31]. It seems that free polymer chains with highermolecular mobility than polyethylene alone are formed in theamorphous phase. This can be attributed to the fact that parts/groups of polymer chains are surrounded by aggregates of silverparticles that limit the interactions of polymer chains at longscale.

Supplementary information about molecular dynamics can befound from the proton spin–lattice relaxation time in the rotatingframe T1qH [32,33]. The composites of PE/Ag-NPs exhibit a signifi-cant decrease of T1qH and this decrease is the same for all Ag-NPcontents. For PE/MWCNT composites, T1qH decreases strictly withincreasing MWCNT filler content.

Table 3Proton spin–lattice relaxation time T1qH and cross-polarization time TCH of PE/MWCNT and PE/Ag-NP composites.

Samples PE/MWCNT PE/Ag

Filler loading (vol.%) 0 0.5 8.5 0 2 5 10 20TCH (ls)a 43.7 43.3 42.5 43.7 121.4 133.7 133.9 274.4TCH long (ls)b 175.3 129.5 174.0 175.3 230.7 250.0 212 274TCH short (ls)b 20.1 16.8 25.6 20.1 8.5 9.1 9.0 9.8T1qH (ms) 61.6 55.0 29.0 61.6 45.4 45.0 42.0 41.1

a Fit with one-component model.b Fit with two-component model.

M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110 107

The T1qH is known to allow the characterization of multi com-ponent material as a bulk with its different domains. T1qH valuesincrease with the degree of crystallinity or with the degree ofordering of the polymer chains. On the contrary, a decrease inT1qH may indicate either an amorphous state or a decrease in thedegree of ordering. In the present case, the main conclusions arethe following: for PE/MWCNT composites a steady decrease isobserved for composites following the addition of MWCNT fillers.This decrease in the T1qH is easily assigned to the paramagneticeffects of the MWCNTs present in these composites. For PE/Ag-NP composites, the values remain in the range of 40–45 ms forall the tested concentrations. When considering the morphologyof these composites given by scanning electron microscopy(Fig. 3), it is clear that the dispersions of MWCNT fillers withinthe polymer matrix are better than those of silver nanoparticles.Therefore the single decrease of T1qH in the case of PE/Ag-NPscould be related to the introduction of inhomogeneity when Ag-NPs were incorporated. Inhomogeneity was mainly due to the for-mation of silver aggregates. Another indication about the differ-ence in homogeneity between the two composites can beprovided by the electrical characterizations, which showed a lowerelectrical percolation threshold (10 vol.%) than that expected forspherical particles (16 vol.%) for the PE/Ag-NP composites [26].On the contrary, the PE/MWCNT composites showed a value ofthe electrical percolation threshold near to the theoretical one [25].

3.4. Conductivity dependence of NMR intensity loss

The conductivity and NMR signal intensity are displayed inFig. 4 for PE/Ag-NP composites. Depending on the concentrationof Ag-NPs, two drops in the NMR signal are observed. The first

Fig. 3. Scanning electron microscopy of (a) PE/MWCNT composite (8.5 vol.%) and (b) PE/Athe PE/Ag one, where aggregates of silver particles are observed (denoted by solid circle

significant decrease occurs for 2 vol.% and the decrease is thenmuch slower from 2 to 10 vol.%. The second important drop hap-pens at the critical concentration for the transition from insulatorto conductive phase (electrical percolation threshold at 10 vol.%).It seems that the NMR signal intensity of the PE carbons is influ-enced not only by the conductive nature of Ag-NPs but also by theirelectrical properties, depending on the volume fraction of conduc-tive Ag-NPs in the matrix. More details of this phenomenon will begiven in the next section by introducing the concept of the skineffect limiting the absorption of the radiofrequency. It should benoticed that in the case of PE/MWCNT, the paramagnetic residuescause more difficulty in investigating this skin effect.

3.5. NMR signal loss and skin effect

Due to the presence of fillers with either paramagnetic or con-ducting properties, the composites were analyzed after dilutionwith silica by mechanical grinding in order to be able to tune theCPMAS probe and to spin the 7 mm diameter rotors to the desiredspinning rate. This technique is based on our previous works onparamagnetic materials [30]. Although the samples were diluted(90 wt.%) in silica as explained previously, a significant disappear-ance of the NMR signal was still observed.

Therefore, the mechanical grinding in silica with 90 wt.%dilution was not sufficient to suppress the signal attenuation,which means that this phenomenon is not due to a macroscopicinteraction or organization between fillers and HDPE involving aweakness of the NMR signal. Indeed, it is worth interpreting asan intrinsic physical phenomenon linked to the presence of con-ductive fillers. For PE/MWCNT the presence of paramagneticMWCNT centers can clearly explain a bleaching of the signal in

g-NP composite (5 vol.%). The PE/MWCNT composite shows more homogeneity thans).

Fig. 4. Normalized NMR-signal intensities (open symbol) and samples dc-conduc-tivities (solid symbol) versus filler concentration of PE/Ag-NP composites (datapoints are linked by solid lines simply to illustrate the evolution of dc conductivityand NMR signal).

Fig. 5. NMR signal intensity versus Ag vol.% for PE/Ag-NP composites. The columnsshow the loss of NMR signal.

Table 4Skin effect values calculated for Ag-NPs at the resonance frequencies of 13C and 1H.

Frequency (Hz) Skin effect (m)

13C 5.00E + 07 8,7217E�61H 2.00E + 08 4,36085E�6

Fig. 6. Schematic representations of the localization of silver nanoparticlesresponsible for the skin effect (orange zones) inside the rotor according to eachmodel. Left: Model 1 – all silver is localized at the surface of the cylinder-rotor.Right: Model 2 – silver nanoparticles are localized at a distance R1 from the centerof the rotor (2D observation).

Table 5Wall size for Ag-NPs at the internal surface of the rotor according to model 1.

Silver nanoparticlescontent (vol.%)

Ag volume insiderotor (m3)

Internalradius-R1 (m)

Wall size(m)

2 2.11E�9 2.99E�3 4.76E�65 4.81E�9 2.98E�3 1.32E�5

10 8.40E�9 2.97E�3 2.45E�520 1.34E�8 2.95E�3 4.03E�5

Fig. 7. Evolution of the normalized NMR signal areas and the wall size (m)calculated according to model 1 as a function of the amount of Ag-NPs.

108 M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110

the proximity of the paramagnetic particles. The loss of NMR signalobserved with silver nanoparticles that are not paramagnetic (seeFig. 5) has been correlated to radiofrequency absorption by the skineffect.

The skin effect created by Ag-NPs has to be calculated at 200and 50 MHz, which are the proton and carbon resonance frequen-cies in a 4.7 T magnetic field. The calculations are performed usingEq. (1) and the values are listed in Table 4:

Skin depth ¼ ds ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2q2pfl0lr

sð1Þ

where q is the bulk resistivity of silver, which has a value of1.5 � 10�8 X m, f is the frequency (Hz), l0 is the vacuum permeabil-ity constant of 4p � 10�7 H m�1, and lr is the relative permeability,which is usually �1 for non-magnetic materials.

From the values listed in the above table, it can be clearly seenthat the skin effects corresponding to the radiofrequency pulsesare compatible with the 3 mm internal radius of the rotor usedfor the NMR measurement as with the size of the silver particlediameter that is in the range of 100 nm. The thickness (wall size)of an equivalent cylinder that can be formed if all the silver parti-cles are localized at the internal surface of the rotor-cylinder (the-oretical model 1) has been calculated as depicted in Fig. 6. The

physical meaning arising from the use of the theoretical model 1is based in its ability to simply predict the values of the thresholdwall size involving a loss of the NMR signal. Therefore, those valuesreported in Table 5 have to be compared with the skin effect valuescalculated in Table 4. With this simple model, we are able to antic-ipate that all composites should exhibit a decrease in NMR signalby partial absorption of the proton or carbon radiofrequencythrough the Ag-NPs except for the sample of PE/Ag-NPs (2 vol.%)for carbon frequency resonance, for which the wall size is smallerthan the 13C skin effect.

Table 6Calculations of the skin effect performed from the signal loss according to model 2.

Ag-NPs(vol.%)

NMRsignal

Corresponding analyzedvolume (m3)

Maskedvolume (m3)

Agthickness(m)

0 1 4.81E�7 2.63E�11 02 0.83 4.02E�7 7.92E�8 1.63E�55 0.81 3.97E�7 8.40E�8 3.65E�5

10 0.79 3.75E�7 1.06E�7 5.70E�520 0.51 2.46E�7 2.35E�7 6.10E�5

Fig. 8. Evolution of NMR signal area and Ag-NP thickness calculated accordingmodel 2 versus Ag-NP (vol.%) content.

M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110 109

to

Fig. 7 presents the NMR signal areas according to the wall size.It can be clearly observed that there is no strict loss in NMR signalintensity with the linear increase of the calculated wall sizeaccording to model 1. Features other than just the conductivenature of Ag-NPs could be set to explain such observations.

Another approach consisting in the calculation of the volume ofmaterial not seen by NMR radiofrequency due to the skin effect hasbeen performed (theoretical model 2). Both observed and maskedvolumes of polyethylene could be calculated using this model,depicted in Fig. 6, in which Ag-NPs are localized at a distance R1from the center of the rotor. In this model the radii R1 and R2correspond to the masked volume of HDPE and the global maskedvolume (HDPE + Ag) respectively. Then the thickness (wall sizeaccording to model 2) of Ag-NPs can be obtained. Table 6 showsthe radius and thickness calculated for all composite contents.

Fig. 8 displays the variation of NMR signal areas and Ag-NPthickness for all PE/Ag-NP composites. As the wall size values arealready above the theoretical 13C skin effect for 2% Ag content,the increase in the amount of Ag-NPs to between 2 and 10 vol.%seems to lead only to an increase of Ag thickness with an almostconstant skin effect due to the conductive nature of the filler. Forthe composition loaded with more than 10 vol.% of Ag-NPs, whichis the value of the electrical percolation threshold in PE/Ag-NPcomposites, it is likely that the silver thickness remains constantwhile a superimposed skin effect is induced by the Ag-NPs, leadingto the second drop of signal intensity. Therefore, the hypothesisinvolving the combination of both the conductive nature of fillersand the electrical percolation network as reasons for the decreasein the NMR signal becomes more reliable.

4. Conclusion

In this study, we demonstrated the potential of solid state NMRas a structural technique to provide complementary informationon the structure at the molecular level of conductive polymer

composites filled with MWCNT and Ag-NPs. By studying themolecular dynamic of polymer chains, signal intensities, andchemical shifts, it was possible to correlate the obtained NMR datawith the morphological analysis. The NMR signal intensitiesdecreased with increases in the amount of conductive fillers. Thisphenomenon was attributed to the skin effect of silver nanoparti-cles in the case of PE/Ag-NPs and was investigated in detail usingsimple models in an attempt to correlate the loss of NMR signaland the electrical properties of the composites.

Acknowledgments

We would like to thank the French Ministry of Higher Educationand Research for PhD grant funding. We acknowledge the ‘‘Centrede Microstructures et d’Analyses, plateform Lyon 1’’ for itsassistance.

References

[1] Fortunati E, D’Angelo F, Martino S, Orlacchio A, Kenny JM, Armentano I. Carbonnanotubes and silver nanoparticles for multifunctional conductive biopolymercomposites. Carbon 2011;49(7):2370–9.

[2] Rybak A, Boiteux G, Melis F, Seytre G. Conductive polymer composites basedon metallic nanofiller as smart materials for current limiting devices. ComposSci Technol 2010;70(2):410–6.

[3] Moniruzzaman M, Winey KI. Polymer nanocomposites containing carbonnanotubes. Macromolecules 2006;39(16):5194–205.

[4] Merhari L. Hybrid nanocomposites for nanotechnology electronic, optical,magnetic and biomedical applications. New York: Springer; 2009.

[5] Noël A, Faucheu J, Rieu M, Viricelle J-P, Bourgeat-Lami E. Tunable architecturefor flexible and highly conductive graphene-polymer composites. Compos SciTechnol 2014;95:82–8.

[6] Bandaru PR. Electrical properties and applications of carbon nanotubestructures. J Nanosci Nanotechnol 2007;7(4/5):1239–67.

[7] Rahman A, Ali I, AlZahrani SM, Eleithy RH. A review of the applications ofnanocarbon polymer composites. NANO 2011;6(3):185–203.

[8] Sanjinés R, Abad MD, Vâju C, Smajda R, Mionic M, Magrez A. Electricalproperties and applications of carbon based nanocomposite materials: anoverview. Surf Coat Technol 2011;206(4):727–33.

[9] Li C, Jiang D, Zeng J, Xing S, Ju S. Electromagnetic transmission characteristicsof composite frequency selective surfaces coated with conductive polymer-silver paste. Compos Sci Technol 2014;90:32–9.

[10] Bauhofer W, Kovacs JZ. A review and analysis of electrical percolation incarbon nanotube polymer composites. Compos Sci Technol2009;69(10):1486–98.

[11] Golnabi H. Carbon nanotube research developments in terms of publishedpapers and patents, synthesis and production. Scientia Iranica2012;19(6):2012–22.

[12] Marcq F, Demont P, Monfraix P, Peigney A, Laurent C, Falat T, et al. Carbonnanotubes and silver flakes filled epoxy resin for new hybrid conductiveadhesives. Microelectron Reliab 2011;51(7):1230–4.

[13] Mdarhri A, Brosseau C, Zaghrioui M, El Aboudi I. Electronic conduction andmicrostructure in polymer composites filled with carbonaceous particles. JAppl Phys 2012;112(3). pp. 034118–034113.

[14] Cecen V, Boudenne A, Ibos L, Novák I, Nógellová Z, Prokeš J, et al. Electrical,mechanical and adhesive properties of ethylene-vinylacetate copolymer (EVA)filled with wollastonite fibers coated by silver. Eur Polymer J2008;44(11):3827–34.

[15] Kwiatkowski J, Whittaker AK. Molecular motion in nanocomposites ofpoly(ethylene oxide) and montmorillonite. J Polym Sci, Part B: Polym Phys2001;39(14):1678–85.

[16] Luo H, Kluppel M, Schneider H. Study of filled SBR elastomers using NMR andmechanical measurements. Macromolecules 2004;37(21):8000–9.

[17] Mirau PA, Lyons M. Intermolecular interactions and dynamics in polymer/C60blends. Macromolecules 2009;43(2):625–9.

[18] Pang W, Fan C, Zhu Q. Solid-state high-resolution 13C NMR study on theinterphase in lamellar semi-crystalline polymers. Eur Polymer J2001;37(12):2425–8.

[19] Umek P, Huskic M, Škapin AS, Florjancic U, Zupancic B, Emri I, et al. Structuraland mechanical properties of polystyrene nanocomposites with 1D titanatenanostructures prepared by an extrusion process. Polym Compos2009;30(9):1318–25.

[20] Powers DS, Vaia RA, Koerner H, Serres J, Mirau PA. NMR characterization of lowhard segment thermoplastic polyurethane/carbon nanofiber composites.Macromolecules 2008;41(12):4290–5.

[21] Bruno Rocha e Silva M, Tavares MIB, Oliveira da Silva E, Neto RPC. Dynamic andstructural evaluation of poly(3-hydroxybutyrate) layered nanocomposites.Polym Testing 2013;32(1):165–74.

[22] Rodrigues T, Tavares M, Soares I, Moreira A. The characterization of high-density polyethylene/organoclay nanocomposites. JOM 2009;61(1):38–41.

110 M. Jouni et al. / Composites Science and Technology 104 (2014) 104–110

[23] Bardet M, Foray MF, Maron S, Goncalves P, Trân QK. Characterization of woodcomponents of Portuguese medieval dugout canoes with high-resolutionsolid-state NMR. Carbohydr Polym 2004;57(4):419–24.

[24] McBrierty V. Solid polymers: a challenge for NMR. Solid State Nucl MagnReson 1997;9(1):21–7.

[25] Jouni M, Boiteux G, Massardier V. New melt mixing polyethylene multiwalledcarbon nanotube nanocomposites with very low electrical percolationthreshold. Polym Adv Technol 2013;24(10):909–15.

[26] Jouni M, Boudenne A, Boiteux G, Massardier V, Garnier B, Serghei A. Electricaland thermal properties of polyethylene/silver nanoparticle composites. PolymCompos 2013;34(5):778–86.

[27] Massiot D, Fayon F, Capron M, King I, Le Calvé S, Alonso B, et al. Modelling one-and two-dimensional solid-state NMR spectra. Magn Reson Chem2002;40(1):70–6.

[28] Li W, Hou L, Chen Z. An NMR Investigation of phase structure and chaindynamics in the polyethylene/montmorillonitenanocomposites. J Nanomater.2013;2013:10.

[29] Bardet M, Gerbaud G, Doan Cu, Giffard M, Hediger S, De Paëpe Gl, et al.Dynamics property recovery of archaeological-wood fibers treated withpolyethylene glycol demonstrated by high-resolution solid-state NMR.Cellulose 2012;19(5):1537–45.

[30] Bardet M, Gerbaud G, Giffard M, Doan C, Hediger S, Pape LL. 13C high-resolution solid-state NMR for structural elucidation of archaeological woods.Prog Nucl Magn Reson Spectrosc 2009;55(3):199–214.

[31] Hoh K-P, Ishida H, Koenig JL. Silicon-29 solid-state nuclear magnetic resonancespectroscopy of composite interfaces. Polym Compos 1990;11(2):121–5.

[32] Arantes TM, Leão KV, Tavares MIB, Ferreira AG, Longo E, Camargo ER. NMRstudy of styrene-butadiene rubber (SBR) and TiO2 nanocomposites. PolymTesting 2009;28(5):490–4.

[33] Motaung TE, Saladino ML, Luyt AS, Chillura Martino DF. The effect of silicananoparticles on the morphology, mechanical properties and thermaldegradation kinetics of polycarbonate. Compos Sci Technol 2012;73:34–9.