1.1.1. Definition and structure

Graphene is a plane of carbon atoms in sp² hybridization. It is the basic constituent of graphite, carbon nanotubes and fullerenes. Figure 1.1 shows a graphene plane.

Graphene can present some structural defects, although its low dimensionality reduces the number of possible imperfections. For instance, the graphene plane may not be strictly composed of hexagons of carbon atoms: these are Stone–Wales defects, where four hexagons are transformed into two pentagons and two heptagons when one of the bonds rotates. Single and double gaps can appear, as well as dislocations, atoms and impurities of substitution. These defects have a strong influence on the physical properties of graphene. In addition, some defects generate dangling bonds that increase the chemical reactivity of graphene (Banhart et al. 2011).

1.1.2. Structures associated with graphene

As will be discussed later in this chapter, a single graphene plane is difficult to obtain and synthesis methods most often result in the stacking of multiple planes, often referred to in the literature as graphene itself. In fact, a specific terminology for carbonaceous materials with a two-dimensional character has been established (Bianco et al. 2013) and the following structures should be distinguished from the single graphene plane:

• – double-wall or triple-wall graphene: structure in free suspension or resting on a substrate, composed of a stack of two or three graphene planes of extended lateral dimension;
• – few-layer graphene (FLG): structure in free suspension or resting on a substrate for a stack of two to five planes;
• – multi-layer graphene (MLG): structure in free suspension or resting on a substrate for a stack of six to 10 planes;
• – graphite nanosheets: two-dimensional graphitic materials with one dimension not exceeding 100 nm;
• – nanoribbon graphene: graphene plane as defined above, with lateral dimensions not exceeding 100 nm and 10 nm, respectively;
• – graphenic carbonaceous materials: include all of the carbonaceous materials made up of sp²-hybridized atoms, like the structures defined above, or the disordered carbons (carbon blacks, activated carbons) and graphite.

1.1.3. Graphene properties

The calculated and measured properties of graphene are unique and remarkable, which explains the interest of the scientific community. Wallace (1947) showed the particular electronic structure of graphene. Thus, graphene is considered as a zero-gap semiconductor. The mobility of charge carriers is high and has been measured up to 230,000 m²·V^–1·s^–1 (Bolotin et al. 2008). This value corresponds to the highest value for a semiconductor. Its thermal conductivity is also remarkable. It reaches 5,300 W·m^–1 ·K^–1 (Rutter et al. 2007), which corresponds to the highest value for a material. In comparison, diamond, which is well known for its ability to evacuate heat, has a thermal conductivity of only 2,000 W·m^–1·K^–1. Its network of sp²-hybridization atoms gives graphene a unique set of mechanical properties. The combination of a hardness equivalent to that of diamond, close to 1 TPa, a very high flexural strength, a high specific stress at break of 48,000 kN·m·kg^–1 and its elongation at break of 25% makes it unique (Lee et al. 2008; Zhu et al. 2010). In comparison, steel has a specific stress at break of 154 kN·m·kg^–1 (Zhu et al. 2010). Graphene has a transparency of 97.7% in the visible spectrum (Rutter et al. 2007; Zhu et al. 2010). Graphene is considered to be the most impermeable material in the world. This is due to the combination of its sp²-hybridization carbon atom lattice with high electron density in its aromatic rings and its structure (covalent bonds and rupture stress) (Berry 2013). Another important property of graphene is that its theoretical specific surface area is 2,630 m²·g^–1 (Ferrari et al. 2015), which is very high for a carbonaceous material.

1.2.1. Graphite exfoliation

1.2.1.3. Graphite exfoliation and functionalization by plasma process in one step

1.2.1.3.1. Plasma by dry-process

Environmentally friendly and economically viable synthesis processes for obtaining pure single-walled graphene with excellent physicochemical properties remain to be discovered. Nevertheless, recent research in the field of plasma physics seems promising (Dey et al. 2016; Levchenko et al. 2016; Giorcelli et al. 2018).

The dry plasma processes used for the preparation or the functionalization of graphene are varied. First, they are differentiated according to the gas pressure, ranging from low pressure processes (a few Pa to 100 Pa) widely known in the literature (Malesevic et al. 2008; Brault et al. 2013; Tsai et al. 2014; Yue et al. 2014) to atmospheric pressure processes in strong development (Subrahmanyam et al. 2009). Low-pressure plasma processes are strongly local thermodynamic non-equilibrium processes with high electron temperatures of up to a few tens of electron volts and a low ionization degree (Samukawa et al. 2012). Plasmas at atmospheric pressure, on the other hand, can exhibit a high degree of ionization, but low electron temperatures (Anders et al. 2009). The pressure difference between the two processes plays a major role in the physical and chemical action that the plasma will have on the etching material. Second, dry processes are differentiated according to the generated plasma regime, which can range from continuous discharges to pulsed discharges and even very high frequency discharges at the electron cyclotron resonance. Plasma arc and jet processes have been mentioned in the literature from the last few years (Wu et al. 2010a; Kim and Suh 2014), mainly because of their ease of implementation and the large quantities of material that can be processed.

1.2.1.3.2. Liquid-phase plasma

Depending on the process, plasma physics is of real interest to prepare or functionalize graphene in the dry process. It is thus possible, for example, to produce liquid-phase graphene nanosheets from asphalt by plasma (Li et al. 2013). On the other hand, the possibility of carrying out both production and functionalization in a single step by means of plasma process requires the removal of scientific and technical barriers, in particular by dissociating purely physical processes from reaction processes. Within this perspective, a first milestone is being achieved at the Institut Jean Lamour in Nancy, by proposing an alternative process based on plasma exfoliation and functionalization of carbon charges in aqueous phase and at atmospheric pressure. In addition to the objective of combining the production of graphene and its functionalization, this new aqueous phase process should also make it possible to avoid the effects of agglomeration and curling of graphene sheets, which are found experimentally and systematically after a synthesis process, before the one enabling the functionalization of graphene. A diagram of the experimental device is proposed in Figure 1.3.

This process operates in the nano- and microsecond-pulsed arc plasma regime on the surface of an aqueous solution loaded with pure graphite particles. A voltage generator coupled to a nanosecond voltage chopper provides a square-wave electrical power signal over the range 0–30 kV and 0–100 kHz, with a pulse duration ranging from 10 to 2000 ns. An electrode immersed in the liquid is connected to the high voltage, so that potential changes are transferred to the surface of the aqueous phase. The upper electrode of the same nature is connected to the ground of the system. The plasma arc discharge is then located between the surface of the solution and the upper electrode. The initiation of the plasma, which is called breakdown, occurs under experimental conditions verifying Paschen’s laws, that is to say when the product between the gas pressure and the distance between the electrodes minimizes the breakdown voltage of the gas. The interest of the high-voltage pulse regime is to promote the exfoliation process of graphite by the ignition of shock waves during plasma breakdown in the air at atmospheric pressure: the carbon planes of the carbon charges located on the surface of the liquid can then undergo a mechanical release. Studies of propagation waves originating from gas bubble formation in a similar case of discharge in liquids indicate that pressures above 10⁷ Pa can be reached locally (Hamdan et al. 2013a). The other advantage of this discharge regime is that it allows for high energy plasma functionalization while having a low energy process, with an average power between 10 and 120 W, depending on the discharge conditions. By imposing only the gas breakdown voltage, it is possible to evolve the process to reach currents of a few amperes. To this end, the aqueous solution will undergo radical changes at the level of its initial intrinsic properties, particularly at its composition level, which will modify its pH and conductivity. This change in liquid composition is due to plasma–liquid interactions, and more specifically to the progressive solvation of plasma-derived species at the liquid surface, such as electrons, ions, and other radicals or energetic species (Lukes et al. 2014; Bruggeman et al. 2016). The evolution in the composition of the medium, its acidification and its progressive rise in temperature are all variable parameters that also promote graphite exfoliation processes on the surface and in the volume of the liquid, as well as reactional mechanisms to functionalize the graphene planes.

1.2.2. Graphite-derived compounds exfoliation

1.2.2.1. Graphite intercalation compounds exfoliation

1.2.2.1.1. Graphite intercalation compounds

Due to its lamellar structure, graphite can accommodate chemical species, called inserts, in its interplanar spaces to give graphite intercalation compounds. Intercalation is a redox reaction that is accompanied by a charge transfer between the graphene planes and the insert; graphite, being amphoteric, can react with electron acceptors and donors. In the well-known example of the KC₈ compound and of alkali metal intercalation in general, the metal acts as an electron donor. It acquires a positive partial charge, while the graphene planes have a negative partial charge.

1.2.2.1.2. Graphite intercalation compounds thermal exfoliation

Thermal exfoliation is most often carried out using intercalation compounds obtained with nitric acid and sulfuric acid. When a violent thermal shock is applied to the intercalation compound, the insert does not have time to diffuse out of the structure and vaporizes directly in the core of the material, causing the graphene planes to spread apart along the axis , due to strong pressure, and a strong increase in volume. The material thus obtained is called exfoliated graphite. It is not very dense and presents an accordion-like structure (Figure 1.4) (Furdin 1998; Inagaki et al. 2004; Afanasov et al. 2010).

Exfoliated graphite can also undergo ultrasonication and grinding steps to yield a material called flat micron graphite (FGM), consisting of single-crystal particles tens of micrometers wide and a few dozen graphene planes thick (Furdin et al. 1991).

1.2.2.1.3. Liquid phase exfoliation of graphite intercalation compounds

The dispersion of a graphite intercalation compound, usually based on alkali metals, in a polar solvent leads to its exfoliation due to charge separation. Electrostatic interactions stabilize the negatively charged graphene planes in suspension. The most widely used compound is KC8 (Catheline et al. 2011, 2012). K(THF)C₂₄ exfoliation has also been reported (Vallés et al. 2008). The exfoliations were initially performed in high-boiling solvents such as NMP but, as with graphite exfoliation, these solvents can be replaced by more easily removed solvents such as tetrahydrofuran (THF). The dispersions are stirred and centrifuged and the resulting particles, of micro-metric size, consist of a stack of one to a few graphene planes (Vallés et al. 2008; Catheline et al. 2011, 2012). The concentrations obtained are of the same order as those of dispersions obtained directly from graphite. Upon the deposition of the dispersions on a substrate, the exposure of the graphene sheets to air returns them to their 0-oxidation state and leads the K^δ+ counterions to oxidize. The oxides are easily removed through washing. The main drawback of this method for obtaining graphene is the extreme sensitivity of the alkali metals and graphite intercalation compounds to air, which must be handled under an inert atmosphere.

1.2.2.2. GO exfoliation

1.2.2.2.1. Graphite oxide

GO is a compound that has been known for many years. This is graphite that has been treated in an oxidizing acid medium. Several historical methods of GO synthesis exist, such as the Hummers method or the Staudenmaier method. The Hummers method consists of treating graphite with sulfuric acid, potassium permanganate and sodium nitrate (Hummers and Offeman 1958). GO, a light-yellow solid, is thus made up of a stack of “graphene” planes carrying various types of oxygenated functions. The oxygenated functions that a carbonaceous material can exhibit are listed in Figure 1.5.

The structure of GO depends on the oxidative treatment (Botas et al. 2013) and the crystalline quality of the starting graphite (Botas et al. 2012). Therefore, it is difficult to establish a typical structure for this type of material and several models have been established (Dreyer et al. 2010). GO mostly contains phenol, carbonyl, carboxyl, ether and lactone functions (Figure 1.5) (Gao et al. 2009). These functions have the effect of doubling the interplanar distance of the compound compared to graphite (Schniepp et al. 2006), and oxidative treatment also leads to the degradation of the sp² lattice.

1.2.2.2.2. GO thermal exfoliation

This technique experimented by Boehm et al. (1962) lead to the first probable observation of graphene and has been recently reutilized with the works on graphene synthesis. When GO is heated rapidly to a temperature of about 1,000°C, the organic functions decompose into CO and CO₂. The pressure thus generated is enough to exfoliate the graphene planes. A black solid composed of crumpled graphene sheets is obtained and after dispersion in NMP or DMF, single-walled graphene with micrometric lateral dimensions can be isolated (Schniepp et al. 2006; McAllister et al. 2007). Oxidative treatment disrupts the sp² lattice of the sheets and thus compromises their properties. Although thermal decomposition of the functions removes much of the oxygen, an ideal graphene structure cannot be restored. Violent heating leads to the additional formation of carbonyl and epoxy groupings, which are thermodynamically very stable and therefore difficult to remove. Furthermore, the removal of the functions leaves sp³-hybridized atoms and atomic gaps within the planes (see Figure 1.6).

1.2.2.2.3. Liquid-phase exfoliation of GO

The oxygenated functions grafted onto the GO planes make them hydrophilic, unlike a purely carbonaceous surface, which is highly hydrophobic. Thus, a simple dispersion in water is sufficient to exfoliate the GO and stabilize particles mostly composed of a single plane (Gomez-Navarro et al. 2007; Stankovich et al. 2007). GO also exfoliates in organic media (Traina and Pegoretti 2012). The material thus obtained is called graphene oxide. It carries the same functions as the parent GO and the carbon/oxygen atomic ratio is between 2 and 3 (Stankovich et al. 2007; Paredes et al. 2008). The characterization of the transport properties indicates a strong degradation of the sp² lattice: graphene oxide is a quasi-insulating material (Gomez-Navarro et al. 2007). In order to restore the properties of graphene oxide, a reduction step is implemented. The most commonly used reducer is hydrazine N2H4. This treatment allows a good part of the oxygen to be eliminated (the carbon/oxygen atomic ratio can rise up to 10) and significantly improves the conduction properties (Gomez-Navarro et al. 2007; Stankovich et al. 2007). However, the charge carrier mobility is 30 cm²·V^–1·s^–1, which still remains much lower than the results obtained with “pure” graphene (Gomez-Navarro et al. 2007). Moreover, as the functionalization is weaker after reduction, the particles become less hydrophilic, tend to reaggregate in aqueous medium and are very difficult to redisperse (Stankovich et al. 2007). Since hydrazine is toxic, several studies have been conducted to replace it. Non-hazardous compounds such as ascorbic acid (vitamin C) are equally effective (Fernandez-Merino et al. 2010). Furthermore, since oxygenated functions are more or less thermally stable, the reduction of graphene oxide can be achieved by thermal reduction at temperatures on the order of 1,000°C (Becerril et al. 2008; Bagri et al. 2010). It has been shown that thermal reduction could lead, at high temperatures, to the formation of particularly stable oxygenated groups that are difficult to eliminate, but it yields good or even better results than chemical reduction, especially with respect to the restoration of the sp²-hybridization lattice and transport properties.

1.3.1. In situ polymerization

In situ polymerization dispersion is a method of nano-composite preparation that can be adapted to a wide range of thermoplastic matrices, such as polyolefins (Fim et al. 2010), polyamides (PAs) (Pan et al. 2000), polystyrenes (PSs) (Chen et al. 2003b), and polymethyl methacrylate (PMMA) (Chen et al. 2003a), but also within thermosetting matrices such as epoxy resins (Miller et al. 2010; Kernina et al. 2019). This method allows for the production of composites from matrices that cannot be used in melt mixing (thermal instability) or dispersed in solution (insoluble polymer). The method is classically carried out in two stages: the first being the mixing of one or more monomers, in powder form or dispersed in a solvent, with the nanofiller, to obtain a homogeneous mixture, then the polymerization of the monomer(s) is carried out to obtain the final nanocomposite. For example, Cromer et al. (2015) performed in situ polymerization of a polypropylene (PP) in the presence of exfoliated graphene fillers. The materials produced by in situ polymerization show better mechanical properties than an equivalent nanocomposite produced by melt mixing, with notably higher elongation at break. Their work highlights the impact that nanofillers can have on polymerization, and the need to modify the conditions in order to maintain the efficient polymerization reaction.

Pol’shchikov et al. (2013) compared two nanocomposites obtained by in situ polymerization via two methods. The nanofillers employed were GNPs produced from oxidized graphite according to the previously mentioned Hummers method. With their first method, the nanofiller, after being heated to 200°C, is placed inside the reactor with dissolved propylene and the metallocene catalyst. With the second method, the nanofiller is first dispersed in toluene and ultrasonicated for 10 or 30 minutes. It is then added to the reactor with propylene and metallocene catalyst. For both methods, the resulting materials were rinsed with a mixture of ethanol and HCl, then washed with alcohol and dried. In both cases, infrared spectroscopy analysis showed that the stereoregularity of the obtained iPP was not influenced by the addition of the nanofiller, but that the use of ultrasound has greatly influenced the dispersion of carbon within the matrix. The two ranges of materials show a close percolation threshold (between 2 and 3 vol%), but the ultrasonically treated materials show higher electrical conductivity and a greater impact of the filler on the mechanical properties with better elongation at break and on the crystallization of the material, with a higher crystallization temperature (Pol’shchikov et al. 2013).

As introduced earlier, it is possible to produce covalent bonds between the nanofiller and the matrix during the polymerization. For example, Huang et al. (2010) performed in situ polymerization of an oxidized PP/graphene nanocomposite with the creation of covalent bonds. The elaboration starts with the preparation of nanofillers via the use of Grignard reagents, in order to obtain a GO nanofiller grafted with Ziegler–Natta catalysts, called TiCl4/(BuMgCl/GO). As presented in Figure 1.7, the obtained filler is then employed for the production of a PP/GO nanocomposite by in situ polymerization. The final nanocomposite shows covalent bonds between the PP polymer matrix and the GO filler. These covalent bonds allow the possible incompatibility between the nanofiller to be bypassed, here polar (GO), and the non-polar polymer matrix, PP, resulting in a better exfoliationand dispersion of the nanofiller. In the work by Huang et al. (2010), an excellent exfoliation of GO fillers is achieved.

It should be noted that the in situ polymerization method can produce a very high level of exfoliation of the carbon planes, especially for intercalated graphite (ICG) or expanded graphite (EG) fillers, by the intercalation of monomer molecules between the carbon planes, and their separation during the polymerization step. This method, which can be called intercalation polymerization, is very similar to the methods used for polymer/clay composites (Ray et al. 2003). The graphene planes are intercalated with monomer molecules in addition to alkali metals in the case of ICGs, or between the accordion planes of EGs (Shioyama et al. 2000). In the case of GO fillers, the distance between oxidized graphene planes, between 0.6 and 0.8 nm, compared to 0.34 nm for a graphite filler, also allows the intercalation of monomer molecules or polymer chains (Jang et al. 2008), particularly in the case of polar polymers such as polyvynilalcohol (PVA) (Ma et al. 2016). The major problem with in situ polymerization development is the complexity of its implementation, but also its high energy consumption, which makes it difficult to apply to high-volume production (Potts et al. 2011; Gulrez et al. 2013).

1.3.2. Intercalation in solution

Solution intercalation processing, like in situ polymerization, provides excellent results for dispersion and exfoliation of a carbon filler within a polymer matrix. The method is applicable to any matrix that can be put into solution in a solvent, such as polyethylenes (PE) (Pang et al. 2010) or PS (Fang et al. 2009). The method simply consists of dispersing the polymer and the nanofiller within the same solvent, and taking advantage of the low viscosity of the mixture to ensure its homogeneity before extracting the solvent. The method is particularly suitable for developing nanocomposites employing an oxidized graphite nanofiller and a water-soluble polymer matrix. Indeed, due to their functionalizations, GO fillers are easily dispersible in water, often by a sonication operation enabling their exfoliation. For instance, Salavagione et al. (2009) produced a PVA/reduced oxidized graphite nanocomposite (RGO often obtained by adding hydrazine) by solution intercalation. The carbon filler employed is an oxidized graphite (GO) developed by the Hummers method. The same method without the reduction step is also performed for the production of PVA/graphite and reduced PVA/oxidized graphite, in order to compare their exfoliation state. Thus, the graphite-filled composites show significant agglomerations in the matrix, while GO and RGO composites show better exfoliation of their fillers, especially for PVA/RGO, with an electrical percolation threshold at 1wt%. The choice of the filler influences the quality of the exfoliation and the dispersion of the final composite, and shows the necessity to adapt the solvent. The importance of the initial dispersion of the filler by ultrasound can also be seen in this case. In an optimal filler/matrix/solvent configuration, it is thus possible to obtain excellent dispersion results, allowing the percolation threshold of the produced nanocomposite to decrease. In some cases it is possible to obtain a percolation threshold lower than 0.5 wt%, as shown by Pang et al. (2010). In their work, a nanocomposite is developed from a high-density polyethylene (HDPE) and an EG nanofiller produced by the Hummers method. Carbon-coated HDPE particles are obtained and shaped by hot pressing. The method of development allows the creation of a percolated lattice, even at low carbon levels during forming. This result is also visible with the evolution of the conductivity represented in Figure 1.8, with an important increase between the materials at 0.005 vol% and 0.076 vol%, and a percolation threshold estimated at 0.070 vol%.

The main limitation of the solution intercalation method comes from the limited number of solvents that can be used, but also from the polymer matrix, which must have a high solubility in order to enable a homogeneous mixing. Moreover, in the case of certain polymers which are difficult to solubilize, such as PP, the use of polar solvents such as xylene is necessary. The intercalation method in solution then requires the evaporation of large quantities of organic solvent, which makes the method not environmentally friendly and hardly adaptable to the industrial production of nanocomposites (Bafana et al. 2017).

1.3.3. Melt mixing

Melt mixing is the most widespread method for the preparation of nanocomposites because of its ease of implementation and the large quantity of materials that can be produced. Moreover, it is the most interesting method from an ecological point of view because it does not involve solvents. Therefore, melt mixing is mainly employed by the industrial sector. There are two main technological routes for this production method: by internal mixing, where the matrix is melted before the addition of the nanofiller, and by extrusion, where the filler can be introduced either at the extrusion inlet with the polymer pellets or during the extrusion process. Because of the great adaptability of an extruder by changing the profile of the conveying screws and the possibility of varying the temperature during mixing, extrusion is the most often favored technology.

Similar to the solution intercalation method which can only be applied to matrices that are soluble in a solvent, the melt mixing method can only be used for matrices that are not too sensitive to thermal degradation. Despite this limitation, it is applicable to many matrices commonly used in industry, such as PP (Kalaitzidou et al. 2007a), PET (Zhang et al. 2010), PA (Canales et al. 2015) and PS (Oxfall et al. 2013). A major shortcoming of this method is a generally higher percolation threshold than with previous methods, mainly due to the fact that the mixture viscosity remains high.

Many authors have proposed methods to improve the dispersion of the filler within the matrix in order to improve its electrical and mechanical properties. The first development parameter that can be easily modified is the mixing speed by screw rotation. Khanam et al. (2016) highlight its importance in the development of a low-density polyethylene/GNP composite. Their results show an improvement in the mechanical properties and electrical conductivity of the nanocomposites, with increasing screw rotation speed due to better dispersion of the nanofiller. However, the authors note that the improvement of the properties is limited when high carbon concentrations are used. Another interesting approach is, as in the case of in situ polymerization or solution intercalation, the use of ultrasound during the production to allow for a better exfoliation and dispersion of the nanofillers. These can either be used during the preparation of the production to ensure the homogeneous mixing of the polymer in powder form with the filler, or directly during the extrusion mixing. He et al. (2018) produced nanocomposites with a GNP mass fraction between 9 and 15 wt% by a single screw extruder coupled with an ultrasonic system, as shown in Figure 1.9. Different configurations were investigated by varying the rotation speed of the extruder and the applied power of the ultrasonic treatment.

These same authors first note an increase in the number and size of the aggregates within the matrix with the increase in carbon concentration. Following the use of ultrasound, an improvement of the dispersion is observed for each concentration. Another important observation is that increasing the power of the ultrasonic treatment makes it possible to reach a better dispersion and even enables the exfoliation of the filler from a certain power, 300 W in the case of their work. The electrical conductivity of the different materials is improved with the increase in the carbon concentration, but also in the case of materials with and without ultrasound treatment. A last interesting result observed by the authors is that an increase in the extruder rotation speed during the ultrasonic treatment decreases the filler dispersion.

A classical variation of the melt mixing preparation makes use of an intermediate step called masterbatch. The method consists of achieving a first polymer/filler mixture with high concentrations (which can exceed 75wt% in nanofiller), then diluting this masterbatch during extrusion to reach the targeted filler concentration. The masterbatch method is indicated for the realization of nanocomposites based on a matrix with a low compatibility with the nanofiller, such as PP, because it is possible to use a derivative close to the matrix during the production of the mixture with high carbon content. For example, Rosehr et al. (2018) employ an atactic PP (aPP) in their works to facilitate the dispersion of their graphite nanofillers within an iPP matrix. Two mixtures were produced by solution mixing in 50 and 80 wt% carbon toluene, respectively. The mixture, after drying, was added to an iPP matrix by extrusion. IPP/graphite melt mixing, without a masterbatch step, was performed in order to study the impact of the aPP on the dispersion. The authors observe a decrease in the crystallization of the nanocomposite with the increase of the fraction of aPP in the mixture, but especially an improvement of the dispersion of the nanofiller, with the disappearance of the large agglomerates present in the nanocomposite without masterbatch. The materials produced from the 50 wt% carbon masterbatch show excellent filler/matrix compatibility due to the aPP remaining on the surface of the fillers. In addition, it limits the reagglomeration of the carbon planes.

In the case of PP, another agent commonly used to improve nanofiller compatibility is maleic anhydride (MAH). MAH is a chemical group that can be grafted to the macromolecular chain of numerous polymers such as PP to provide new functionalities. In the case of carbon nanofiller dispersion, the polarity of the MAH grouping allows for improved dispersion, whether the blends are obtained in solution (Li et al. 2018), by melt mixing (Shanks and Cerezo 2012), or via a masterbatch (Li et al. 2017). Similar work was conducted on polyethylene by Shen et al. (2005). The addition of MAH to the nanocomposite elaboration opens up another elaboration possibility, allowing improved dispersion of the nanofiller through the creation of covalent bonds. One of the methods that can be used is based on the high reactivity of MAH with nitrogenous groups, like amines and amides. This reaction makes it possible, for example, to provide new functionalities to the polymer (Cui and Paul 2007), improve the mixing of a copolymer (Colbeaux et al. 2005) or perform matrix cross-linking (Létoffé et al. 2019). Hsiao et al. (2011) use this reactivity in a way that transforms their GO fillers into chemical cross-linking nodes, according to the reaction scheme described in Figure 1.10. A mixing without grafting operation is also performed by melt mixing for comparison. The cross-linked composites show better dispersion, better compatibility of the nanofiller with the iPP matrix, as well as an increase in the crystallinity ratio and melting temperature of the material. Several variations of this method exist in the literature, employing different amino molecules to create the matrix/filler covalent bonds (Yuan et al. 2014; Li et al. 2018). Bao et al. (2014), on the other hand, use octadecylamine (ODA) and demonstrate a 100% improvement in impact strength and 30% improvement in tensile strength for a PP.

1.3.4. Comparison of development methods

Several authors have performed the comparison between the different methods described above (Stankovich et al. 2006; Kalaitzidou et al. 2007a; Kim et al. 2010b; Potts et al. 2011; Sengupta et al. 2011; Gulrez et al. 2013). The in situ polymerization method appears to be the most interesting when taking the percolation threshold and final material properties into consideration. Cromer et al. (2015) made several iPP/GNP nanocomposites via two different methods: by in situ polymerization and by melt mixing. The range of materials produced by melt mixing surprisingly exhibits better dispersion, but finitely less attractive mechanical and electrical properties than its equivalent by in situ polymerization.

Similarly, Kalaitzidou et al. (2007b, 2008) have compared a conventional solution intercalation method, a filler-coated matrix method, as well as a melt-mixing method. In their case, the solution-based method, and especially the coating variant, showed better electrical properties, with a lower percolation threshold and better mechanical properties. The difference in properties is explained by a difference in morphology and dispersion, in particular by the presence of agglomerates within the melt-mixing matrix and dispersed charges in the coating-based method. The coating method has the advantage of preserving the morphology of the GNPs in the final composite, allowing extremely low percolation thresholds to be obtained. Although apparently less interesting than the other methods, the melt-mixing method remains one of the most preferred since it is the simplest to set up and does not require any solvent during the preparation. Several variants of the different methods, such as the use of masterbatch or filler-based matrix coating prior to processing, are good ways to improve the desired properties. Table 1.1 summarizes the advantages and disadvantages of the three main processing methods discussed above.

1.4.1. Analysis of the material morphology

Because of the important link between the quality of the nanofiller dispersion and the mechanical and thermoelectric properties of the final nanocomposite, the analysis of the morphology of the material is a determining step of its characterization. The classical methods used for the analysis of nanofiller dispersion are transmission electron microscopy (TEM), scanning electron microscopy (SEM) and X-ray diffraction at large and small angles (WAXS and SAXS). The use of the X-ray method shows the presence of large agglomerates by different diffraction peaks, but also, in the case of graphite-derived fillers, the evolution of the distances between carbon sheets, especially for GO and EG fillers, can be studied (Kim and Macosko 2009). In the case of the dispersion of graphite fillers and its derivatives, an interesting parallel can be drawn with silicate nanofillers, which have already been widely studied in the development of nanocomposites (Paul and Robeson 2008; Ray and Okamoto 2008; Bhattacharya 2016). In both cases, the dispersion and exfoliation of the sheets are highly dependent on the preparation conditions and can result in different morphologies, as illustrated in Figure 1.11.

For example, Huang et al. (2010) employ both SEM and TEM imaging techniques to characterize the morphology of their PP/GO composites produced by in situ polymerization. Their preparation method based on the surface grafting of their GOs with Ziegler–Natta catalysts enables the surface polymerization of carbon planes promoting their exfoliation. Using these methods, it is possible to highlight the presence of nanofillers and characterize their states of dispersion, but also estimate the number of planes composing the GO fillers after preparation, as well as appreciate the quality of the adhesion of the matrix to the filler. It should be noted, however, that some morphologies are difficult to observe, especially in the case of particularly exfoliated fillers, namely in the form of perfectly isolated sheets in the matrix, either by TEM or X-ray (Paul and Robeson 2008; Kim et al. 2010b). In addition, the cryomicrotomy cutting necessary for sampling can prove complicated depending on the matrix employed.

1.4.2. Influence of nanofillers on semi-crystalline microstructures

In the case of a semicrystalline matrix, the carbon filler can impact the crystallinity rate, the thicknesses of crystalline lamella and/or the crystallization speed. In the case of PP, carbon planes behave as nucleating agents, but limit macromolecular mobility and thus crystal lamella growth (Kalaitzidou et al. 2008; Yuan et al. 2014; He et al. 2018). Filler compatibility with the matrix, as well as processing and dispersion conditions play a very important role. Therefore, in their work comparing in situ polymerization and melt-mixing processing methods, Cromer et al. (2015) show that in both situations an increase in the melting temperature of their iPPs results in a higher crystallinity rate as a function of the increasing mass fraction of carbon particles. However, this result is twice as large for materials obtained by melt mixing. In contrast, Bafana et al. (2017) show a decrease in the crystallinity rate and only a very small variation in the melting temperature for their materials produced by a solution-mixing method. On the other hand, Hsiao et al. (2011), in their work on nanofiller grafting, show a strong difference between PP/TRGO materials produced by simply dispersing their nanofillers, and PP/PP-g-TRGO materials, in which the matrix and the nanofiller exhibit covalent bonds and thus better compatibility. This difference in behavior highlights the importance of the compatibility between the matrix and the filler in the production of nanocomposites.

The crystallization rate of a matrix is also strongly impacted by the insertion of fillers in a semi-crystalline matrix. In the case of PP, several authors have studied the effect of graphite fillers and its derivatives on the crystallization rate, as well as on the final morphology of the material depending on the amount, but also the size of the nanofiller (Kalaitzidou et al. 2008; Zhao et al. 2014; Jun et al. 2018a). Zhao et al. (2014) observe that the presence of graphite, even at low concentration, allows surface, but especially edge nucleation of carbon sheets at temperatures higher than pure materials. The preferential localization at the edge of the sheets can be explained by the presence of a greater number of defects in the structure of the filler, making a better nucleation possible. The authors also observe a decrease in the size of the crystalline structures due to their greater number and the space limited by the presence of graphite. These results are confirmed in the work of Kalaitzidou et al. (2008) on the crystallization of a PP charged with different graphite fillers. The crystallization rate increases significantly with the addition of graphite: it can triple between a virgin PP and a PP/GNP at 0.01 wt%, and sixfold for concentrations up to 1 wt%. This increase of the crystallization rate, but also of the number of spherulites developing around the nanofillers, implies a decrease in their final sizes.

The effect of the size of the graphite filler in terms of diameter is also studied by Kalaitzidou et al. (2008). Their optical analyses show that at equivalent loading rates, the materials with the largest loads crystallize as larger spherulites, while exhibiting close crystallinity rates. The X-ray diffraction analysis shows a significant difference between the materials containing a filler of 1 μm and 15 μm diameter. While increasing the concentration of both fillers impacts the α-phase of iPP in the same way – an increase in the intensity of the diffraction peaks of the plane family (00l) is observed – a difference appears for the β-phase confirmed by Jun et al. (2018b). The 1 μm diameter fillers at low concentrations (0.01 wt%) enable the crystallization of the material under its β phase, but not at higher concentrations (1 wt%). On the contrary, the addition of the 15 μm diameter filler systematically induces a β-phase even at higher concentrations (e.g. 10 wt%). This difference comes from the large number of nucleation points created by the 1 μm diameter fillers, which by rapidly occupying the available space do not allow the formation of β-spherules. The appearance of the β phase is also indicated by other studies (Shanks and Cerezo 2012; Ruy and Shanmugharaj 2014; Ahmad et al. 2017; He et al. 2018; Rosehr et al. 2018). In general, the nucleating effect is higher for smaller fillers because they provide a larger specific surface area of contact with the matrix and limit the movement of polymer chains less (Vallés et al. 2014).

1.4.3. Influence of nanofillers on mechanical properties

One of the main reasons for adding a carbon or another nanofiller to a polymer matrix is to improve its mechanical properties. Therefore, the effect of carbon nanofillers on the mechanical properties of polymer matrices is studied in many formulations, and as in the case of the influence on the microstructure, the effects vary greatly depending on the chosen configuration. For example, exfoliated graphene fillers appear to be more suitable for mechanical reinforcements than black or carbon nanotube fillers (Kim et al. 2010a; Bhattacharya 2014). Similarly, the compatibility between filler and matrix can greatly influence the mechanical properties of the materials. Polar polymers such as PVA, PMMA and PAN, but also epoxy matrices, are known for large increases in their Young’s moduli and glass transition temperatures with the addition of carbon nanofillers, especially GO, due to the non-covalent bonds that can be created between the matrix and the chemical groups present on the surface of the carbon particles (Ramanathan et al. 2008; Rafiee et al. 2009; Ren et al. 2018). In general, carbon nanofillers induce an increase in the Young modulus and tensile strength of the material, parallel to the rate under load. Thus, it is generally accepted that the nanofiller limits the movement of the macromolecular chains of the material, leading to its stiffening. This evolution is generally achieved to the detriment of the elongation at break of the material and depends on the adhesion between filler and matrix.

These trends are observed in the case of PP-based nanocomposites. For example, Jun et al. (2018a) show that the addition of large-diameter GNPs (>150 μm) at 1.3 wt% allows the increase of the Young modulus of their PP matrices from 514 to 540 MPa, but strongly decreases its elongation at break, ranging from 908% to only 17%. Some authors also note a loss of ductility of the nanocomposite (Song et al. 2011; Yuan et al. 2014; Rosehr et al. 2018). This decrease can be explained by the tendency of the fillers to agglomerate, making decohesion with the matrix easier. Zhao et al. (2010) compare this condition to filler “mechanical percolation”. Improved filler dispersion and exfoliation yields better mechanical properties. Cromer et al. (2015), in their comparisons of in situ polymerization and melt-mixing methods, note that at equivalent filler rates a better ductility of their matrices is produced by the first method, for which a better dispersion state was observed. Similarly, Kalaitzidou et al. (2007b, 2008) observe an improvement of the mechanical properties of their composites via their method of elaboration by coating, also resulting from better filler dispersion.

In addition to the dispersion quality, the size of the fillers has an influence on the final mechanical properties of the nanocomposite. As previously stated, it influences PP crystallization, in particular by promoting the development of the β-crystalline phase which, although the latter presents a better impact strength than the α-phase, shows lower mechanical tensile strength. More generally, Young et al. (2018) explain that the stress transfer between the matrix and the nanofiller and the Young modulus of the nanocomposites are globally independent of the intrinsic properties of the nanofiller, but mainly depend on their form factor and compatibility with the matrix. Jun et al. (2018b) thus compare four types of GNPs with different diameters: 2 μm (C300), 5 μm (M5), 25 μm (M25) and 150 μm (H100), for mass concentrations ranging from 1 to 10 wt%. The results of their mechanical analyses are presented in Figure 1.12. The range of nanocomposites filled with H100 GNPs shows an increase in Young’s modulus with the carbon concentration, but a low tensile strength, while the other three ranges show an improvement of both parameters, but all at the expense of the elongation at break. This evolution in H100 materials is explained by the large size of the carbon sheets that fold and form large agglomerates. On the contrary, the materials based on the smaller filler sizes show an improvement in both their Young’s moduli and their tensile strength, with a maximum reached for the C300 filler, the smallest in this study. Therefore, decreasing the size of the filler makes it possible, at equivalent loading rates, to increase the contact area and thus achieve a better stress transfer between the matrix and the filler (Kuvardina et al. 2013; Liang et al. 2016).

Finally, the last parameter that can influence the mechanical properties is the compatibility of the matrix with the surface of the filler. A compatibility agent is then used as an intermediary to yield a better mechanical filler transfer.

Rosehr et al. (2018) employ atactic PP chains as compatibilizing agents on the surface of GNPs when performing a masterbatch. After mixing within an iPP matrix, the resulting materials exhibit tensile strength equivalent to a composite produced without a masterbatch step, but show higher elongation at break, as shown in Figure 1.13. It is interesting to note that the elongation at break is all the higher when the aPP mass fraction is large. The presence of aPP chains on the surface of the carbon ensures a better transfer of the stress and limits the phenomenon of filler/matrix decohesion within the nanocomposite.

Other authors use grafting of different functional chemical groups such as maleic anhydride onto the macromolecular chains. The resulting chemical adhesion induces a net improvement of the dispersion of the filler within its matrix, but also allows the significant improvement of its mechanical properties (Colbeaux et al. 2005; Li et al. 2017). The main drawback of these methods originates from the impact of these groups and their grafting operations on the properties of the pristine matrix. Other types of grafting are listed in the literature, and those inducing covalent bonds with the filler provide significant improvement in mechanical properties (Hsiao et al. 2011; Ryu and Shanmugharaj 2014).

1.4.4. Influence of nanofillers on electrical properties

Regardless of their shapes, the choice of carbon nanofillers in the development of a nanocomposite is mainly due to their excellent electrical properties compared to other fillers (Bhattacharya 2016). However, in order to avoid having to add them in too large quantities, which could adversely affect other properties of the final material, the best method to increase conductivity is to improve dispersion by reducing the agglomeration phenomenon. This strong impact of dispersion is explained by the phenomenon of percolation of the charge within the matrix by the creation of a necessary and essential conductive lattice (Kirkpatrick 1973; Lux 1993).

1.4.4.1. The role of percolation

During the preparation of a nanocomposite, three states are possible, according to the rate of charge and the dispersion of the latter. At a low loading rate, the fillers are separated by the insulating matrix, the conductivity of the material is then close to that of the initial matrix. As the load rate increases, the filler-to-filler contact becomes more frequent, until a threshold concentration is reached, called the percolation threshold, from which the charges form a continuous lattice in the material, called a percolated lattice. The electrical conductivity of the material then presents a strong increase. Beyond this threshold, the conductivity of the material stabilizes at values close to the conductivity of the filler. In addition to this contact conductivity, the phenomenon of “tunneling” occurs for sufficiently close charges, for distances less than 10 nm, enabling electrical conductivity before the contact percolated lattice is formed (Al-Saleh and Sundararaj 2009; Stauffert et al. 2009). Figure 1.14 shows the different possible configurations. In case of poor filler distribution and dispersion (configuration (a)), and in case of good distribution but poor dispersion (configuration (b)), no conductive lattice is formed. In case of good distribution and dispersion (configuration (d)), the distance between the charges can prevent the formation of the network. At low filler concentration, the optimal configuration is therefore configuration (c), with good dispersion, but a distribution of fillers enabling the formation of the continuous network.

Many parameters are involved in the percolation phenomenon, whether it is the polymer matrix used, its compatibility with the filler, the preparation method, or the form factor of the filler. In their review on the production of polymer/graphene nanocomposites, Potts et al. (2011) compare the lowest electrical percolation thresholds obtained with several polymer/filler configurations. Their observations are summarized in Table 1.2. Significant variations in the percolation threshold are observed, down to values below 0.2 wt% for an oxidized polyethylene/graphene composite, as seen for the solution-based preparation by Pang et al. (2010).

Polymer matrix	Nanofiller	Percolation threshold (wt%)	Reference
Epoxy	Fonct. EG	1.0	(Miller et al. 2010)
Nylon-6	GO	0.5	(Du et al. 2010)
Polyaniline (doped)	GNP	0.7	(Wu et al. 2010b)
Polycarbonate	TEGO	0.3	(Khanam et al. 2016)
Polyethylene	RG-O	0.15	(Pang et al. 2010)
Polyethylene terephthalate	TEGO	1.0	(Zhang et al. 2010)
Polymethyl methacrylate	GNP	0.7	(Chen et al. 2003a)
Polypropylene	GNP	0.7	(Kalaitzidou et al. 2007b)
Polystyrene	Fonct. GO	0.2	(Stankovich et al. 2006)
Polyvinyl alcohol	RG-O	0.5	(Salavagione et al. 2009)
Polyvinyl chloride	GNP	1.4	(Vadukumpully et al. 2011)
Polyvinylidene fluoride	TEGO	2.0	(Ansari et al. 2009)
Polyurethane	TEGO	0.6	(Kim et al. 2010b)

1.4.4.2. Evolution of electrical properties

As in the case of the influence on the microstructure and on the mechanical properties, many parameters influence the electrical properties of the nanocomposite, such as the size of the charge, its dispersion, the compatibility with the matrix, but also in the case of semi-crystalline materials, the crystallinity of the material. It is possible, depending on the configuration, to reach electrical conductivities of the order of S·cm^–1 (Kalaitzidou et al. 2007b; Jun et al. 2018a).

He et al. (2018) show that the use of ultrasound during the preparation of their materials allows an increase in the conductivity of the materials, resulting in a better dispersion. This difference in percolation, according to the quality of the dispersion, is particularly visible in the work by Kalaitzidou et al. (2008), who compare the levels of dispersion obtained according to the methods of elaboration in solution and by melt mixing. The latter shows a percolation threshold above 5 vol%, while the solution method gives a percolation between 0 and 3 vol% (Figure 1.15).

Jun et al. (2018a) show a difference in the electrical properties of their PP/GNP materials depending on the size of the introduced fillers. The first observation made by the authors is the lack of change in electrical conductivity with the addition of their smallest filler (2 μm diameter). The larger the filler diameter, the lower the percolation threshold, which is in agreement with previous results on the filler size/percolation threshold relation proposed by Wu et al. (2013) and Jun et al. (2015). It is interesting to note that this evolution is to the advantage of large diameter fillers and is in contrast to the previously observed evolutions related to microstructure and mechanical properties, both of which are more favorable to small-sized fillers. A trade-off must then be found between electrical and mechanical properties.

Kalaitzidou et al. (2008) show that the cooling rate of the material during its preparation strongly impacts the percolation threshold of the nanocomposite. Indeed, the percolation threshold is four to five times lower for materials cooled slowly. This difference is explained at the microstructural level. During fast cooling, a large number of small spherulites form on the surface of the carbon films and block the formation of the electrical percolation lattice, whereas slow cooling allows for the formation of large spherulites favorable to percolation.

A similar finding can be made for nanocomposites utilizing compatibilizing agents that allow covalent-type adhesion between the matrix and the filler. The better dispersion obtained via this method is responsible for a decrease in the percolation threshold, as well as an increase in conductivity (Hsiao et al. 2011; Ryu and Shanmugharaj 2014; Li et al. 2017).

1.4.5. Evolution of the thermal resistance

One of the known consequences of the addition of a carbon nanofiller within a polymer matrix is the increase of matrix thermal stability. This can be observed in air thermogravimetric analysis (ATG) by an increase in the degradation temperature of the material with the filler ratio. For example, Bafana et al. (2017) obtain an increase in the degradation temperature of their iPP with the addition of graphene (Figure 1.16).

This increase in material degradation temperature is not linear with increasing loading rate, but increases significantly from low values and then reaches an asymptote (Song et al. 2011; Kuvardina et al. 2013; Zhao et al. 2014; Ahmad et al. 2017; Ren et al. 2018). This peculiarity can be explained by the role of nanofillers in the permeability to gases, more specifically to the oxygen of the nanocomposite. Indeed, the graphene sheets, if devoid of defects, present an impermeability to gaseous molecules and thus make it possible to limit their diffusion within, and through, the polymer matrix (Bunch et al. 2008). The resulting nanocomposites then become good candidates as protective materials against corrosion, in the form of an impermeable film (Yang et al. 2013). At equivalent filler ratio, the effect is more important for small diameter fillers (Figure 1.17) due to the increase in the interaction surface between the filler and the matrix (Vadukumpully et al. 2011; Beuguel et al. 2018). Improved compatibility between matrix and filler implies better thermal energy diffusion in the matrix and thus also improves the thermal resistance of the composite. On the other hand, Hsiao et al. (2011) find that the addition of 0.1wt% of a TRGO filler induces an increase in the thermal degradation temperature of an iPP. This property is only promoted if the filler is grafted to the matrix.

Finally, we note that the large difference between the degradation temperatures of the polymer and the nanofiller enables, after preparation, a verification of the mass percentage remaining after complete degradation of the polymer matrix.