Polypropylene Carbon Nanotubes Nanocomposites: Combined Influence of Block Copolymer Compatibilizer and Melt Annealing on Electrical Properties

We study the influence of melt annealing and the presence of a block copolymer compatibilizer on the electrical properties of polypropylene carbon nanotubes (CNT) nanocomposites from the DC limit to microwave frequencies and link it to the morphological details. We show that the compatibilizer concentration controls three types of morphologies: separate CNT agglomerates, a network of well dispersed but interconnected CNT, and individualized but separate nanotubes. This explains why conductivity reaches an optimum over the whole frequency range at a low compatibilizer concentration. We model the corresponding structures by a semiquantitative schematic equivalent electrical circuit. A key outcome of the work is the understanding and control of dispersion mechanisms in order to optimize the electrical performances for efficient EMI shielding depending on the targeted frequency range.


Introduction
The ever expanding use of electrically insulating polymers as enclosure materials for electrical, electronic, and high frequency (HF) wireless communication devices exacerbates electrostatic dissipation and electromagnetic interference (EMI) issues but at the same time opens up new opportunities for original solutions.
For bulk or surface DC conductivity, the main challenge is to conceive materials that retain high mechanical performance combined with adequate electrical conductivity.This naturally leads to (nano)composites of a dielectric matrix and a well dispersed conductive filler, preferably with high aspect ratio, providing electrical percolation at low concentration.Polymer-carbon nanotubes (CNT) nanocomposites are clearly attractive from that standpoint if dispersion can be controlled [1][2][3].It is important to realize that the contact resistance between CNT dominates the conductivity of percolation networks in DC and low frequency regimes.This implies that while optimal dispersion of the CNT is favorable for conductivity, the contact distance and resistance between them must be minimized to allow conduction by electron hopping [4][5][6][7].Insulating irreversible coating on the CNT is hence detrimental.
EMI shielding is nowadays a very challenging issue because electronic devices are becoming ever more compact and integrated, which increases spurious coupling interferences either by direct transmission or through multiple reflections.This leads to a renewed interest for electromagnetic absorber materials, which simultaneously reduce the reflection and transmission of the signal.Besides metamaterials designed to absorb specific frequencies by resonance [8], composite-based materials containing conductive inclusions, for example, metal powders, ferrites, carbon, and conjugated polymers, have already been proposed as electromagnetic (EM) absorbers [9][10][11][12][13][14][15][16][17][18][19].Again, well dispersed polymer-CNT nanocomposites (possibly foamed, see [20,21]) are attractive for broadband absorption in the GHz frequency range because the high aspect ratio of CNT allows reaching sufficient conductivity at low filler content and hence the elusive combination of high conductivity and low dielectric constant, which is essential for effective absorption [5,[21][22][23][24].HF conductivity of polymer composites follows different rules than DC conductivity.Above a critical frequency, electron hopping is no longer the dominant mechanism but conductivity results from Resistive-Capacitive (RC) coupling between filler and matrix, which can be described by equivalent circuits [5,[25][26][27].EMI shielding CNT-based polymer nanocomposites based on, for example, polycarbonate, polycaprolactone, polypropylene and polystyrene have already been described [20,21,[28][29][30].They still mostly work by reflection, not absorption.Recently, effective EM absorbing structures based on polymer-CNT nanocomposites have been described [21,[31][32][33].
Polyolefin-CNT nanocomposites, either polyethylene or polypropylene (PP) based, have attracted a lot of academic and industrial interest due to their unique combination of cost effectiveness and properties [21,24,[34][35][36].However, the dispersion of pristine CNT in polyolefins by melt-mixing is usually poor, due to unfavorable interactions between the polymer chains and the CNT as well as strong Van der Waals attraction between the CNT themselves [37,38].This is particularly true for PP, often resulting in poor conductivity and mechanical properties of the composites.To ensure a good load transfer between the CNT and the polymer matrix as well as a good dispersion of the nanofiller, CNT modifications by organic molecules, including polymer chains, can be used.For example, some authors have proposed a covalent modification route to help CNT dispersion in PP [21,36,39,40].However, grafting techniques have the disadvantage of being cumbersome and detrimental to true electrical contacts.Physical coating of the CNT by selective adsorption is a preferable strategy for several reasons: first, it preserves all intrinsic features of the nanotubes; second, it provides efficient surface modification with ease and cost effectiveness; third, it is potentially reversible, which can be of importance to reduce contact resistance between the CNT.The effectiveness of this selective adsorption strategy for improved CNT dispersion or phase/interfacial selectivity has been demonstrated by our research team for acrylateethylene copolymers, polyamides, and polyethylene already [37,[41][42][43].One key result of these studies is that the melting sequence of the polymers plays an important role for the final dispersion and morphology: the first polymer to contact the nanotubes during melting has the opportunity to strongly adsorb by noncovalent interaction.Overall, the control of DC as well as HF conductivity of CNT-polymer nanocomposites results from a complex balance between CNT dispersion and electrical networking.At low frequency, the contact resistance between the percolating CNT is the critical parameter while, at high frequency, the details of the morphology control the RC coupling.It is known that melt annealing can allow electrical contacts between the CNT to strengthen under the influence of their mutual Van der Waals attraction [37,38].This is called secondary agglomeration.However, the mutual attraction between CNT will be stronger in typical situations where they are difficult to disperse (high value of the Hamaker constant).At the opposite, when CNT disperse effectively in a polymer matrix, it is often because their surface has been modified with an insulating organic coating, which is detrimental to electrical contacts.Very few studies have been devoted to a systematic analysis of the combined influence of CNT-polymer compatibility and melt annealing on DC and HF conductivity.On the other hand, there is little literature about composites for absorption below 1 GHz; EMI shielding was observed around 200 MHz using carbon nanotubes in poly-methyl-methacrylate [44].Up to 30 dB of shielding was obtained.However these performances result from a combination of reflection and absorption rather than pure absorption.Regarding polypropylene composites interesting properties were observed at 800 MHz in [45].In [46] up to 15 dB shielding effectiveness was obtained with polypropylene loaded with carbon fibers.In [47] 25 dB shielding efficiency was obtained; absorption was observed at X-band in polypropylene foams loaded with carbon fibers.Carbon nanotubes are also efficient for absorption above 1 GHz.Indeed their high aspect ratio promotes the formation of a conductive path at a low content [33].Mechanical behavior of absorbers is also of prime interest in airborne applications.Therefore some solutions were investigated combining CNT composite and honeycomb metallic matrix in order to improve resistance to impact [48].
In this work, we use block copolymer compatibilizer (hydrogenated styrene-isoprene-styrene triblock: SEPS) combined with melt annealing to modify the dispersion and electrical contacts of pristine multiwall CNT in a PP matrix.The styrene blocks of SEPS have an affinity for CNT because of their polarisability and potential for pi-pi stacking.The EP blocks on the other hand have an affinity for the PP phase.SEPS block copolymers can hence act as effective compatibilizers, as will be demonstrated below.Varying the SEPS concentration at constant CNT load allows us to systematically vary the nanotubes dispersion, mutual attraction in the matrix, and the resulting electrical properties.We also study the influence of annealing the nanocomposites in the melt on their DC and HF conductivity.The electrical properties are studied over a broad frequency range thanks to the combination of two different techniques, operating in the 0.01 Hz to 1 MHz and 1 GHz to 67 GHz ranges, respectively.A key outcome of this work is the understanding and control of dispersion mechanisms in order to optimize electrical performances for efficient EMI shielding, depending on the targeted frequency range.

Raw Materials.
The CNT used in this study are provided by the company FutureCarbon.They are homogeneous multiwall tubes obtained by the CVD method, with 10-20 concentric graphene layers and an average diameter of ∼20 nm.Purity is 98 w/w%.SEPS copolymer Hybrar 7125 is produced by Kuraray Co. Ltd by hydrogenation of the corresponding

Blending and Compression Molding.
A twin-screw minicompounder (DSM Microextruder 15 cm 3 ) is used to compound the nanocomposites.For all melt blending experiments, the conditions are as follows: 250 rpm, 10 minutes compounding at 220 ∘ C. All components of the blends are added simultaneously in the hopper after dry-mixing.For further characterization of electrical properties, the compounded samples are compression-molded at 190 ∘ C for 1, 10, and 20 min, respectively, in a Fontijne press followed by quick cooling.

Nanocomposite Samples.
All nanocomposite samples contain 2 wt% CNT.In order to enable an exhaustive analysis of the correlation between the structure and electrical properties, we systematically vary the PP/SEPS weight ratio of the polymer matrix.The composition and nomenclature of the samples (except for the pure matrix references) are reported in Table 1.For the same composition, annealing time can range from 1 to 20 min.

Transmission Electron Microscopy (TEM).
The specimens are cut with the help of a Reichert Microtome.Ultrathin sections of approximately 70-100 nm in thickness are cut using a cryodiamond knife (Diatome, Switzerland) and collected on 400 mesh copper grids.CNT dispersion in the nanocomposites is investigated with the help of a LEO 922 Transmission Electron Microscope operating at 200 kV or a FEI Tecnai 12 operating at 120 kV.To increase contrast between the PP and SEPS phases in blends, some ultrathin sections are exposed to RuO4 vapor for 30 min or to OsO4 for 2 min.The styrene blocks are selectively stained by RuO4 and thus show up as darker phases on the TEM images.The OsO4 vapor reacts to stain the residual carbon-carbon double bonds present in SEPS.

Electrical Measurements.
For the frequency range between 0.01 Hz and 1 MHz, a Novocontrol Alpha-A High Performance Frequency Analyser with a Agilent Dielectric Test Fixture is used.The samples are placed between two electrodes to measure the resistance and capacitance, using a parallel equivalent circuit for modeling the behavior.The relative dielectric constant   and electrical conductivity  of the sample under study are obtained from the measured capacitance  and resistance  according to the expressions   = (/  ) ⋅ (/) and  = /( ⋅ ), where  is the sample area,  the sample thickness, and   the permittivity of vacuum.
For higher frequencies (between 1 GHz and 67 GHz), a Vector Network Analyser (model N5247A PNA-X Network Analyser from Agilent) is used with a Anristu Model 3680 V test fixture in order to measure the  parameters of the microstrip line configuration and extract the dielectric constant and electrical conductivity of the sample, according to the method described in [48].

Results
In nanocomposites, the electrical properties are highly correlated to the nanostructure of the CNT network.Indeed, dispersing conductive nanoparticles in a polymer matrix results in a modification of the dielectric constant and electrical conductivity of the resulting composite.A fine tuning of these parameters is required to maximize absorption of electrical signals in the composite, resulting in improved EMI shielding.In this work, TEM imaging (effective at the nm-m scale) and electrical measurements (which probe the structure at the mm-cm scale) are used as complementary techniques to characterize the dispersion of CNT in the matrices.1-6 present TEM pictures of the nanocomposites tested in this work.They show both the quality of the CNT dispersion and the morphology of the PP-SEPS mixtures.Most images correspond to composites annealed for 10 minutes in the melt during compression molding of the previously compounded samples (see Experimental Part), after OsO4 or RuO4 staining.Exceptions are indicated in the caption of the figures.Except for samples with no or low amount of SEPS, which are not affected by annealing time (example in Figure 1), the morphology of the blends evolves upon annealing time until a stable and reproducible structure is observed after approximately 10 min, as can be inferred from the absence of significant evolution between 10 and 20 min (example in Figure 2).Melt annealing for 10 min hence allows sufficient macromolecular interdiffusion to reach overall morphological and compositional homogeneity.Images of samples annealed for only 1 min are shown only for samples containing no or limited to small amounts of SEPS, the morphology of which is unaffected by annealing.

Morphology of PP/SEPS/CNT Nanocomposites. Figures
In Figure 3 the first and the second line correspond to microtomes of the nanocomposites stained by OsO4 and RuO4, respectively.On the RuO4 stained images, bright unstained areas corresponding to PP and dark stained dots corresponding to the styrene blocks of SEPS can be seen.On the other hand, on the OsO4 stained images, the dark dots correspond to the residual carbon-carbon double bonds      in the EP phase of SEPS while the other phases (S in SEPS and PP) remain unstained.The images first show the good overall compatibility of SEPS with PP (see Figures 3(b) and 3(c)).This is due to the presence of EP blocks in SEPS, which are compatible with polyolefins, in particular PP.Second, Figure 3 highlights the very favorable influence of SEPS on the dispersion of CNT in PP.In pure PP (Figure 3(a)), the CNT form big lumps.They are impossible to disperse correctly due to their strong mutual attraction and initial entangled state (from synthesis).Addition of a small quantity of SEPS (5%, see Figure 3(b)) dramatically improves CNT dispersion, which remains of course excellent also at higher SEPS concentrations (Figures 3(c) and 3(d)).
The strong affinity of the styrene blocks of SEPS for the CNT is confirmed in Figure 4 and is consistent with literature results [24].At very high SEPS concentration, the styrene blocks entirely cover the CNT and tend to separate them, hence acting as an insulating interlayer.
The detrimental influence of high SEPS concentration can be inferred from the analysis of Figure 5. Indeed, when PP50 is stained by RuO4 (Figure 5(a)) strong darkening around the CNT indicates the presence of SEPS styrene blocks surrounding the nanotubes.Similarly, when PP50 is stained by OsO4, a light grey halo around the CNT reveals SEPS is entirely coating the CNT (Figure 5(c)).Hence, in this sample, the nanotubes are physically separated, which is good from the standpoint of CNT dispersion but bad for conductivity.Such halo is absent in PP0 (Figure 5(b)), confirming it is due to SEPS coating the nanotubes.
Figure 6 shows the same systems at a smaller magnification to highlight the overall state of dispersion of the CNT.The chosen staining procedure has been optimized for optimal clarity of the structure imaging.Cartoon insets clarify the evolution from separate lumps in pure PP to fully individualized CNT in pure SEPS with a clear evolution at intermediate PP/SEPS ratios, that is, tight clusters linked by loose connections at 5% SEPS and loose clusters with extensive connections in the 48% SEPS case.

AC Dielectric Constant and Conductivity PP/SEPS
Nanocomposites with 2 wt% CNT.At constant CNT loading in the nanocomposite, the balance between CNT agglomerates and the network of well dispersed CNT, and their respective arrangement, modifies the electrical conductivity and dielectric constant.Figures 7 and 8 show the conductivity and dielectric constant values obtained for the reference samples and nanocomposites annealed for 10 min.Although there is a frequency gap between the ranges covered by the two measurement methods, the results are very consistent as the curves from one range can be smoothly extrapolated to the other range.The conductivity and dielectric constant for the reference samples (darker curves) are lower than the corresponding values for the composites.In a pure dielectric polymer, the AC conductivity is mainly due to molecular relaxation at high frequency and DC conductivity is vanishingly small.The PP98, PP93, PP50, and PP0 composites (samples are referenced by their PP content at constant 2% CNT loading, see Experimental Part) have a conductivity comprised between 10 −13 S/m and 1 S/m in the explored frequency range, while the dielectric constant value ranges from 3 to 1000.A conductivity plateau appears at low frequency for the PP98 and PP95 samples, evolving to a power law (straight line in double log representation) at higher frequencies, with a transition zone in between.For the other two composite samples (PP50 and PP0), a power law is observed over the entire frequency range.The highest conductivity and dielectric constants are obtained at 5% SEPS (PP93) whereas either high PP (PP98) or high SEPS (PP50, SEPS98) content is less favorable.The differences are the most spectacular at low frequencies.
When SEPS is absent (PP98) or poorly dispersed (large agglomerates of SEPS inside PP are observed in PP50 after only 1 min annealing), some micrometric agglomerates of CNT are always present.

Discussion
The great challenge of correlating dielectric spectroscopy measurements with TEM micrographs is caused by the very different representative volumes of the two methods.Indeed, dielectric measurements give information about a volume around 0.1 cm 3 (global information) versus only a few m 3 for TEM images (very local information).The dispersion and the percolation threshold can however be linked through the correlation between microscopy observations and measured electrical parameters (electrical conductivity and dielectric constant).The dielectric constant and the AC conductivity depend upon different processes: fluctuations of the dipoles, the propagation of mobile carriers (electrons, holes, or ions diffusion), and the separation of charges at the interfaces.This last mechanism leads to an additional polarization which takes place at inner boundary layers (Maxwell-Wagner-Sillars polarization).This contribution can be larger than the dielectric response due to molecular fluctuations [49].Whereas CNT are lumped as big separate agglomerates in pure PP, a small SEPS concentration of 5% strongly enhances their dispersion and a network between the CNT agglomerates starts to appear (Figures 6(a)-6(c)).When the SEPS concentration is further increased, the strong affinity between the SEPS and the CNT results in a layer of insulating macromolecules all around the nanoparticles (see Figure 4), which partially destroys electrical contacts between them and hence decreases conductivity (Figure 6(d)).To support this interpretation of the experimental results, the conduction mechanism can be represented by the schematic and corresponding equivalent electrical circuit illustrated in Figure 9. Low frequency conduction arises from the network of CNT that are in direct contact.Conductivity is further enhanced at higher frequencies, thanks to capacitive coupling between nontouching CNT agglomerates.The different structures can be modeled as the electrical components of an equivalent electrical circuit.The fluctuations of the molecular dipoles in the hosting polymer matrix correspond to a capacitance; the propagation of the mobile carriers and the separation of the charges at the interface can be represented by different values of a resistance.At low frequency, high conductivity can result from low resistivity CNT agglomerates, provided that they are linked by a sufficient percolating network path.
The equivalent circuit has the following impedance: where  CNT is the CNT resistance,  CNT is the capacitance existing between closely adjacent but isolated/nontouching CNT conductive agglomerates, R( dipolar ) is the capacitance associated with molecular fluctuations/molecular dipoles in the hosting polymer matrix, and tan  is its loss tangent factor while  network is the resistance accounting for the conductive network existing inside the polymer matrix due to existing physical contacts between conductive fillers: CNT having resistance  CNT either interconnected or creating conductive paths between isolated agglomerates having resistance  agglo .The dielectric polarization losses of the polymer matrix can have a significant influence and are accounted for by an imaginary part of the equivalent dipolar capacitance, while factors  and  account for the relative volumetric fractions of interconnected and isolated CNT agglomerates in the hosting matrix.Predictions of the frequency dependence of the conductivity using the electrical circuit model are shown in Figure 10 for two limiting cases of well dispersed CNT and large agglomerates, respectively.The model's parameters are given in Table 2.
The low frequency measurements are highly dependent on the resistive component of the composite: the resistance is higher for well dispersed CNT (PP50 and PP0) than for networks containing agglomerates (PP93).This is probably due to higher contact resistance between the nanotubes owing to an adsorbed insulating SEPS layer (see Figure 4).On the other hand, the high frequency behavior mainly depends on capacitive coupling and follows a power law versus frequency, independently of the microstructure, but the structure of the network and the CNT dispersion strongly influences the conductivity close to the percolation rate.The physics of the material is mainly related to the different properties of each component and the internal structure of the composite.For higher rates of CNT, the network looks like a global agglomerate of entangled CNT.And the current can then pass across many paths, increasing the electrical conductivity and the crossover frequency.At lower rates of CNT, the number of conductive particles is too low to form an electrical network, independently of the network structure, and then the composite is an insulator material.9 for two limiting cases of well dispersed CNT and large agglomerates, respectively.Model's parameters corresponding to each case are given in Table 2.

Absorption
The electrical properties influence the electromagnetic absorption.The samples PP0 and PP50 have a low dielectric constant independent of the frequency; and the electrical conductivity shows a power law behavior as a function of the frequency.The samples PP0 and PP93 show a different behavior at low frequency: the dielectric constant and electrical conductivity dramatically increase below 100 kHz.This behavior directly impacts the absorption: the coupled values (electrical conductivity and dielectric constant) for an optimal absorption depend consequently on the frequency; hence the best material for absorption changes if the frequency is modified.The absorption can be estimated as a function of the dielectric constant and conductivity, frequency, and thickness of the sample, according to the formalism described in [48,50].
6.1.1 Hz-100 MHz Frequency Range.On Figure 11 the absorption levels are very low (less than 1%).The absorption is globally higher for PP93 and increases for frequencies above 100 MHz.The absorption of PP98 also increases above 100 kHz.The reflection decreases due to the reduction of the dielectric constant while the transmission decreases due to the increase of the electrical conductivity resulting in an increase of the absorption.

100 MHz-67 GHz.
The electrical conductivities show a power law behavior and are almost parallel.The range goes from PP0 to PP98 and to PP93 (which has the highest conductivity).The sample PP50 has a particular behavior and thus is not included in the first step of comparison.The high values of conductivity imply a significant reduction of the transmission factor which radically decreases the transmission level and impacts also the reflection level (less significative increase).Figure 12 shows the absorption rate as a function of frequency for sample thickness of 2 mm.The optimal couple is 3-0.01 to 10 S/m.The sample PP93 is the best material for this frequency range.Increasing the frequency does not influence the order of materials having the best absorption for this thickness (PP0, PP98, and PP93) which is the same as the measured conductivity.

Conclusion
We have studied the link between the dispersion of multiwall CNT inside a melt compounded PP matrix and the corresponding electrical properties over a broad frequency range as a function of melt annealing time and the concentration of an SEPS block copolymer compatibilizer.The annealing time is an important process parameter which significantly influences the overall morphology and conductivity for samples around 50/50 PP/SEPS ratio.In such case, melt annealing for 10 min allows sufficient macromolecular interdiffusion to reach overall morphological and compositional homogeneity.On the other hand, samples containing only a very small concentration of the minor phase are unaffected by annealing.The low as well as high frequency  conductivity reaches an optimum at small concentration of compatibilizer.This is due to a balance between the presence of highly conducting but separate CNT agglomerates and of a conducting path network made of individualized CNT.The balance between the two is controlled by the compatibilizer concentration.If the latter is too low, the CNT agglomerates remain separate and the composite does not conduct.At the opposite, when the compatibilizer concentration is too high, the CNT form an extensive network but the contact resistance between them reduces conductivity due to the presence an insulating adsorbed layer.The schematic structure observed as a function of compatibilizer concentration (after sufficient annealing time) can be semiquantitatively modeled by an equivalent electrical circuit which shows that resistive effects dominate at low frequency while capacitive ones dominate at high frequency.

Figure 7 :Figure 8 :
Figure 7: (a) Electrical conductivity in the 1 Hz-67 GHz frequency range.The lowest conductivities are measured for the pure SEPS and PP matrices.With 2% wt CNT, the conductivities range from 10 −13 to 10 −4 S/m at low frequencies, depending on PP content.The highest conductivity is obtained at 5% SEPS (PP93).(b) Electrical conductivity in the 1.7 GHz-67 GHz frequency range.

Figure 9 :Figure 10 :
Figure 9: Schematic of the CNT dispersion mechanisms inside a polymer matrix: CNT agglomerates closely separated by polymer, capacitance associated with dipolar/molecular fluctuations of free/unfilled polymer areas, and DC/low frequency conduction paths mechanisms: agglomerates connected by isolated CNT, and network of interconnected well dispersed CNT.Corresponding equivalent circuit includes (a) a resistance and a capacitance representing the closely spaced agglomerates and their capacitive coupling through polymer separating them, (b) a capacitance associated with dipolar motion, and (c) a resistance modeling the network of interconnected conductive fillers (agglomerates and/or dispersed CNT).

Figure 11 :
Figure 11: Predictions of absorption for thickness of sample = 2 mm, in frequency range 1 Hz-100 MHz.

Figure 12 :
Figure 12: Predictions of absorption for thickness of sample = 2 mm, in frequency range 100 MHz-67 GHz.

Table 1 :
Compounded samples compositions and nomenclature.It has a high vinyl content (1-2 addition of the isoprene monomer during anionic polymerization of the isoprene block is favored over 1-4 addition) and has a melt flow rate of 4 g/10 min at 230 ∘ C under 2.16 kg load.Isotactic polypropylene HF420FB is purchased from Borealis.It has a melting point of 161 ∘ C and a melt flow rate of 19 g/10 min at 230 ∘ C under 2.16 kg load.