Dispersion of carbon nanofillers in mono- and biphasic polymer matrices : thermodynamic vs. kinetic and influence on high frequency electrical properties

Electronic devices are omnipresent in our daily life. There is therefore a need for electromagnetic interference shielding devices. In this context, the interest of researchers for the development of conducting polymer, based on the incorporation of conductive fillers is increasing exponentially. Among conductive fillers, carbon nanotubes (CNT) and, more recently, graphene have attracted tremendous attention due to their exceptional mechanical, electrical and thermal properties. The main key challenge remains however the dispersion of CNT in selected polymers. The objective of this thesis is to answer some fundamental questions related to the dispersion and distribution of CNT and other nanofillers in single-phase polymer matrices and immiscible polymer blends. First the key aspects controlling the dispersion of CNT in different polymer matrices are described. They are from both thermodynamic and kinetic origins. The enthalpic contribution is dominated by van der Waals (VDW) interactions between as-produced CNT. To minimize these interactions, the difference between the dielectric constants and refractive indexes of the CNT and the matrix has to be minimized. The kinetic parameters are described by the mixing energy, which must overcome the strong attractive interactions between CNT. Increasing mixing energy, by increasing mixing rotation speed and time and decreasing compounding temperature, leads to improved dispersion. However, the electrical properties are reduced. The achievement of a very good dispersion state can thus be harmful to the formation of a conductive network. Second, as the best electrical properties are not always obtained for the best dispersion states, the formation of percolating networks in ethylene-acrylate copolymers (EA) and the resulting Radio Frequency (RF) electrical properties are analyzed using rheological and electrical characterizations. The initially well dispersed CNT form interconnected agglomerates when annealing the sample in the melt state. The modulus, G’, and the conductivity are thus enhanced. This phenomenon is called secondary agglomeration and is time and temperature dependent. For CNT content lower than the percolation threshold (1 and 2wt.-% CNT), the formation of the interconnected network occurs faster when increasing temperature. Above the percolation threshold (5wt.-% CNT), the formation of the network is independent of temperature. The network structure is also more quickly achieved when increasing the CNT content. In order to elaborate conducting materials with decreased conducting filler content, immiscible polymer blends can be used to exploit the concepts of double or triple percolation. Indeed, when dispersing CNT in a blend of two immiscible polymers, they can either locate in one phase or in both or at the interface. It is therefore essential first to understand the parameters that govern the localization of CNT in immiscible polymer blends. The CNT equilibrium position is determined by thermodynamics and more precisely by the evaluation of the wetting coefficient. According to this coefficient, at the equilibrium, the CNT should locate in one phase (called preferred phase) or at the interface. The mixing strategy and the relative viscosities have also a high effect on the final morphology of the blends. When the carbon nanotubes are first dispersed in the preferred phase, whatever the relative viscosities, they stay well dispersed in this phase. On the contrary, when they are first dispersed in the less preferred phase, they migrate towards the preferred phase. However, the total transfer can be hindered if the preferred phase presents a very high viscosity compared to the less preferred phase or if the less preferred phase is adsorbed on the CNT surface. In these cases, migration in the preferred phase is partially or totally prevented and at least a fraction of the CNT is observed at the interface of the phases. As a consequence of the interfacial confinement of CNT, the morphology of blends presenting a sea-island morphology is stabilized due to the presence of a deformable barrier at the interface, preventing coalescence of the dispersed droplets. This stabilization is observed for EA / polyamide (PA) blend for long mixing time (at least 60 min) and low CNT content (0.5 wt.-% CNT). The interfacial localization observed for the blend EA/PA6 is next exploited to elaborate triple continuous systems. The effect of CNT on co-continuous morphology and the resulting electrical properties are studied. Two different mixing strategies are used: predispersion of CNT in EA and predispersion of CNT in PA6. Both mixing strategies lead to the formation of a conductive pathway at the interface. However, for the predispersion in EA, the conductivity is dominated by agglomerates. The co-continuous domain is wider in the presence of CNT than in the absence of CNT, whatever the mixing strategy. The blends exhibiting a triple continuous structure however exhibit low RF conductivities. This can be partially explained by the presence of a high number of sub-inclusions surrounded by CNT, which are therefore lost for the elaboration of a conductive network, and possibly also by the high anisotropy of the structure. In comparison, a blend exhibiting a sea-island morphology with large interconnected dispersed droplets present a much higher conductivity. Surprisingly, the triple continuity is therefore not ideal for the elaboration of conductive composites. Note finally that in the presence of CNT, the co-continuous blends, exhibit improved stability. The stabilization is more efficient for systems with predispersion of CNT in EA because of the better CNT interfacial confinement. For predispersion in PA6, a coarsening of the morphology is observed but the co-continuous morphology is preserved. However, the electrical properties are not significantly affected. A full knowledge of the system modification resulting from polymer degradation is however required to analyze rigorously the stability. Next, the concepts developed for CNT are extended to graphite-based platelets. As for CNT, the dispersion of graphite nanoplatelets (GNP) in single-phase polymer matrices is correlated to the minimization of the Hamaker constant. Furthermore, the dispersion of high amounts of GNP (>15wt.-%) followed by compression highly orients the platelets. A resonant effect in the RF electrical properties is associated to this orientation. The system behaves like an electrical circuit consisting of a resistor, an inductor, and a capacitor, connected in series (RLC circuit). On another hand, when dispersing graphite-base platelets in immiscible polymer blends (EA/PA6), depending on the mixing strategy, the platelets either partly migrate towards the interface, when predispersed in the less preferred phase (EA) or remain dispersed in the preferred phase (PA6). However, the PA6 domains are not so well surrounded by platelets than by CNT. Finally, we study the combination of CNT and silver nanoparticles (Ag-NP) in a single polymer matrix (PA6) and immiscible polymer blends (EA/PA6) and the resulting electrical properties. Both fillers are homogeneously dispersed in PA6. An affinity between the fillers is observed. However, it does not enhance the electrical properties at high frequencies. When dispersed in immiscible polymer blends (EA/PA6), both fillers migrate towards the interface. The concentration of the fillers leads to improved electrical properties at high frequencies for co-continuous polymer blends and to enhanced local DC conductivity for blends with a sea-island structure.(FSA 3) -- UCL, 201