Polymer-layered Silicate Nano Composites. Preparation, Properties

Materials Science and Engineering, 28 (2000) 1±63

Polymer-layered silicate nanocomposites: preparation, properties and uses of a new class of materials Michael Alexandre, Philippe Dubois*

Laboratory of Polymeric and Composite Materials, University of Mons-Hainaut, 20 Place du Parc, B-7000 Mons, Belgium Accepted 20 March 2000

Abstract This review aims at reporting on very recent developments in syntheses, properties and (future) applications of polymer-layered silicate nanocomposites. This new type of materials, based on smectite clays usually rendered hydrophobic through ionic exchange of the sodium interlayer cation with an onium cation, may be prepared via various synthetic routes comprising exfoliation adsorption, in situ intercalative polymerization and melt intercalation. The whole range of polymer matrices is covered, i.e. thermoplastics, thermosets and elastomers. Two types of structure may be obtained, namely intercalated nanocomposites where the polymer chains are sandwiched in between silicate layers and exfoliated nanocomposites where the separated, individual silicate layers are more or less uniformly dispersed in the polymer matrix. This new family of materials exhibits enhanced properties at very low filler level, usually inferior to 5 wt.%, such as increased Young's modulus and storage modulus, increase in thermal stability and gas barrier properties and good flame retardancy. # 2000 Elsevier Science S.A. All rights reserved. Keywords: Layered silicate nanocomposites; Intercalative polymerization; Melt intercalation; Exfoliation±adsorption; Mechanical properties; Thermal stability

1. Introduction Manufacturers fill polymers with particles in order to improve the stiffness and the toughness of the materials, to enhance their barrier properties, to enhance their resistance to fire and ignition or Abbreviations: AFM, atomic force microscopy; AIBN, N,N0 -azobis(isobutyronitrile); ALA, aminolauric acid; APP, ammonium polyphosphate; BDMA, benzyldimethylamine; BTFA, boron trifluoride monomethylamine; CEC, cation exchange capacity; DGEBA, diglycidyl ether of bisphenol A; DMA, dynamic mechanical analysis; DSC, differential scanning calorimetry; EDX, energy dispersive X-ray; EVA, ethylene vinyl acetate copolymer; FTIR, Fourier transform infrared spectroscopy; HDPE, high density poly(ethylene); HPMC, hydroxypropylmethylcellulose; HRR, heat release rate; MAO, methylaluminoxane; MMT, montmorillonite; NBR, nitrile rubber; NMA, nadic methyl anhydride; PAA, poly(acrylic acid); PAN, poly(acrylonitrile); PANI, poly(aniline); PBD, poly(butadiene); PCL, poly(e-caprolactone); PDDA, poly(dimethyldiallylammonium); PDMS, poly(dimethylsiloxane); PEO, poly(ethylene oxide); PFT, polymerization-filling technique; PI, poly(imide); PLA, poly(lactide); PP, poly(propylene); PP-MA, maleic anhydride modified poly(propylene); PP-OH, hydroxyl modified poly(propylene); PPV, poly(p-phenylenevinylene); PS, poly(styrene); PS3Br, poly(3-bromostyrene); PVA, poly(vinyl acetate); PVCH, poly(vinylcyclohexane); PVOH, poly(vinyl alcohol); PVP, poly(2-vinyl pyridine); PVPyr, poly(vinylpyrrolidone); PXDMS, poly(p-xylenylene dimethylsulfonium bromide); SBS, symmetric(styrene±butadiene±styrene) block copolymer; SEC, size exclusion chromatography; TEM, transmission electron microscopy; TEOS, tetraethylorthosilicate; THF, tetrahydrofuran; XRD, X-ray diffraction * Corresponding author. Tel.: ‡32-65-373481; fax: ‡32-65-373484. E-mail address: [email protected] (P. Dubois) 0927-796X/00/$ ± see front matter # 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 7 - 7 9 6 X ( 0 0 ) 0 0 0 1 2 - 7

2

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 1 Example of layered host crystals susceptible to intercalation by a polymer Chemical nature

Examples

Element Metal chalcogenides Carbon oxides Metal phosphates Clays and layered silicates

Graphite [8] (PbS)1.18(TiS2)2 [9], MoS2 [10] Graphite oxide [11,12] Zr(HPO4) [13] Montmorillonite, hectorite, saponite, fluoromica, fluorohectorite, vermiculite, kaolinite, magadiite, . . . M6Al2(OH)16CO3
nH2O; MˆMg [14], Zn [15]

Layered double hydroxides

simply to reduce cost. Addition of particulate fillers sometimes imparts drawbacks to the resulting composites such as brittleness or opacity. Nanocomposites are a new class of composites, that are particle-filled polymers for which at least one dimension of the dispersed particles is in the nanometer range. One can distinguish three types of nanocomposites, depending on how many dimensions of the dispersed particles are in the nanometer range. When the three dimensions are in the order of nanometers, we are dealing with isodimensional nanoparticles, such as spherical silica nanoparticles obtained by in situ sol±gel methods [1,2] or by polymerization promoted directly from their surface [3], but also can include semiconductor nanoclusters [4] and others [2]. When two dimensions are in the nanometer scale and the third is larger, forming an elongated structure, we speak about nanotubes or whiskers as, for example, carbon nanotubes [5] or cellulose whiskers [6,7] which are extensively studied as reinforcing nanofillers yielding materials with exceptional properties. The third type of nanocomposites is characterized by only one dimension in the nanometer range. In this case the filler is present in the form of sheets of one to a few nanometer thick to hundreds to thousands nanometers long. This family of composites can be gathered under the name of polymer-layered crystal nanocomposites, and their study will constitute the main object of this contribution. These materials are almost exclusively obtained by the intercalation of the polymer (or a monomer subsequently polymerized) inside the galleries of layered host crystals. There is a wide variety of both synthetic and natural crystalline fillers that are able, under specific conditions, to intercalate a polymer. Table 1 presents a non-exhaustive list of possible layered host crystals. Amongst all the potential nanocomposite precursors, those based on clay and layered silicates have been more widely investigated probably because the starting clay materials are easily available and because their intercalation chemistry has been studied for a long time [16,17]. Owing to the nanometer-size particles obtained by dispersion, these nanocomposites exhibit markedly improved mechanical, thermal, optical and physico-chemical properties when compared with the pure polymer or conventional (microscale) composites as firstly demonstrated by Kojima and coworkers [18] for nylon±clay nanocomposites. Improvements can include, for example, increased moduli, strength and heat resistance, decreased gas permeability and flammability. The aim of this report is to review the different techniques used to obtain polymer-layered silicates nanocomposites and the improved properties that those materials can display. 2. Generalities 2.1. Structure of layered silicates The layered silicates commonly used in nanocomposites belong to the structural family known as the 2:1 phyllosilicates. Their crystal lattice consists of two-dimensional layers where a central

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 1. Structure of 2:1 phyllosilicates (reproduced from [19] with permission).

octahedral sheet of alumina or magnesia is fused to two external silica tetrahedron by the tip so that the oxygen ions of the octahedral sheet do also belong to the tetrahedral sheets. The layer thickness Ê to several microns is around 1 nm and the lateral dimensions of these layers may vary from 300 A and even larger depending on the particular silicate. These layers organize themselves to form stacks with a regular van der Walls gap in between them called the interlayer or the gallery. Isomorphic substitution within the layers (for example, Al3‡ replaced by Mg2‡ or by Fe2‡, or Mg2‡ replaced by Li‡) generates negative charges that are counterbalanced by alkali or alkaline earth cations situated in the interlayer. As the forces that hold the stacks together are relatively weak, the intercalation of small molecules between the layers is easy [16]. In order to render these hydrophilic phyllosilicates more organophilic, the hydrated cations of the interlayer can be exchanged with cationic surfactants such as alkylammonium or alkylphosphonium (onium). The modified clay (or organoclay) being organophilic, its surface energy is lowered and is more compatible with organic polymers. These polymers may be able to intercalate within the galleries, under well defined experimental conditions as will be reported about in Section 3. Montmorillonite, hectorite and saponite are the most commonly used layered silicates. Their structure is given in Fig. 1 [19] and their chemical formula are shown in Table 2. This type of clay is characterized by a moderate negative surface charge (known as the cation exchange capacity, CEC and expressed in meq/100 g). The charge of the layer is not locally constant as it varies from layer to layer and must rather be considered as an average value over the whole crystal. Proportionally, even if a small part of the charge balancing cations is located on the external crystallite surface, the majority of these exchangeable cations is located inside the galleries. When the hydrated cations are ion-exchanged with organic cations such as more bulky alkyammoniums, it usually results in a larger interlayer spacing. Table 2 Chemical structure of commonly used 2:1 phyllosilicatesa 2:1 Phyllosilicate

General formula

Montmorillonite Hectorite Saponite

Mx(Al4ÿxMgx)Si8O20(OH)4 Mx(Mg6ÿxLix)Si8O20(OH)4 MxMg6(Si8ÿxAlx)O20(OH)4

a

Mˆmonovalent cation; xˆdegree of isomorphous substitution (between 0.5 and 1.3).

3

4

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 2. Alkyl chain aggregation in layered silicates: (a) lateral monolayer; (b) lateral bilayer; (c) paraffin-type monolayer and (d) paraffin-type bilayer (reproduced from [21] with permission).

In order to describe the structure of the interlayer in organoclays, one has to know that, as the negative charge originates in the silicate layer, the cationic head group of the alkylammonium molecule preferentially resides at the layer surface, leaving the organic tail radiating away from the surface. In a given temperature range, two parameters then define the equilibrium layer spacing: the cation exchange capacity of the layered silicate, driving the packing of the chains, and the chain length of organic tail(s). According to X-ray diffraction (XRD) data, the organic chains have been long thought to lie either parallel to the silicate layer, forming mono or bilayers or, depending on the packing density and the chain length, to radiate away from the surface, forming mono or even bimolecular tilted `paraffinic' arrangement [20] as shown in Fig. 2. A more realistic description has been proposed by Vaia et al. [21], based on FTIR experiments. By monitoring frequency shifts of the asymmetric CH2 stretching and bending vibrations, they found that the intercalated chains exist in states with varying degrees of order. In general, as the interlayer packing density or the chain length decreases (or the temperature increases), the intercalated chains adopt a more disordered, liquid-like structure resulting from an increase in the gauche/trans conformer ratio. When the available surface area per molecule is within a certain range, the chains are not completely disordered but retain some orientational order similar to that in the liquid crystalline state (Fig. 3). This interpretation has been recently confirmed by molecular dynamics simulations where a strong layering behavior with a disordered liquid-like arrangement has been found, that can evolve towards a more ordered arrangement by increasing the chain length [22]. As the chain length

Fig. 3. Alkyl chain aggregation models: (a) short alkyl chains: isolated molecules, lateral monolayer; (b) intermediate chain lengths: in-plane disorder and interdigitation to form quasi bilayers and (c) longer chain length: increased interlayer order, liquid crystalline-type environment (reproduced from [21] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 4. Scheme of different types of composite arising from the interaction of layered silicates and polymers: (a) phaseseparated microcomposite; (b) intercalated nanocomposite and (c) exfoliated nanocomposite.

increases, the interlayer structure appears to evolve in a stepwise fashion, from a disordered to more ordered monolayer then `jumping' to a more disordered pseudo-bilayer. 2.2. Nanocomposite structures Depending on the nature of the components used (layered silicate, organic cation and polymer matrix) and the method of preparation, three main types of composites may be obtained when a layered clay is associated with a polymer (Fig. 4). When the polymer is unable to intercalate between the silicate sheets, a phase separated composite (Fig. 4a) is obtained, whose properties stay in the same range as traditional microcomposites. Beyond this classical family of composites, two types of nanocomposites can be recovered. Intercalated structure (Fig. 4b) in which a single (and sometimes more than one) extended polymer chain is intercalated between the silicate layers resulting in a well ordered multilayer morphology built up with alternating polymeric and inorganic layers. When the silicate layers are completely and uniformly dispersed in a continuous polymer matrix, an exfoliated or delaminated structure is obtained (Fig. 4c). Two complementary techniques are used in order to characterize those structures. XRD is used to identify intercalated structures. In such nanocomposites, the repetitive multilayer structure is well preserved, allowing the interlayer spacing to be determined. The intercalation of the polymer chains usually increases the interlayer spacing, in comparison with the spacing of the organoclay used (Fig. 5), leading to a shift of the diffraction peak towards lower angle values (angle and layer spacing values being related through the Bragg's relation: lˆ2d sin y, where l corresponds to the wave length of the X-ray radiation used in the diffraction experiment, d the spacing between diffractional lattice planes and y is the measured diffraction angle or glancing angle). As far as exfoliated structure is concerned, no more diffraction peaks are visible in the XRD diffractograms either because of a much too large spacing between the layers (i.e. exceeding 8 nm in the case of ordered exfoliated structure) or because the nanocomposite does not present ordering

5

6

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 5. XRD patterns of: (a) phase separated microcomposite (organo-modified fluorohectorite in a HDPE matrix); (b) intercalated nanocomposite (same organomodified fluorohectorite in a PS matrix) and (c) exfoliated nanocomposite (the same organo-modified fluorohectorite in a silicone rubber matrix) (reproduced from [19] with permission).

anymore. In the latter case, transmission electronic spectroscopy (TEM) is used to characterize the nanocomposite morphology. Fig. 6 shows the TEM micrographs obtained for an intercalated and an exfoliated nanocomposite. Besides these two well defined structures, other intermediate organizations can exist presenting both intercalation and exfoliation. In this case, a broadening of the diffraction peak is often observed and one must rely on TEM observation to define the overall structure.

Fig. 6. TEM micrographs of poly(styrene)-based nanocomposites: (a) intercalated nanocomposite (reproduced from [60] with permission) and (b) exfoliated nanocomposite (reproduced from [61] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

3. Nanocomposite preparation Several strategies have been considered to prepare polymer-layered silicate nanocomposites. They include four main processes [23]:  Exfoliation±adsorption: The layered silicate is exfoliated into single layers using a solvent in which the polymer (or a prepolymer in case of insoluble polymers such as polyimide) is soluble. It is well known that such layered silicates, owing to the weak forces that stack the layers together can be easily dispersed in an adequate solvent. The polymer then adsorbs onto the delaminated sheets and when the solvent is evaporated (or the mixture precipitated), the sheets reassemble, sandwiching the polymer to form, in the best case, an ordered multilayer structure. Under this process are also gathered the nanocomposites obtained through emulsion polymerization where the layered silicate is dispersed in the aqueous phase.  In situ intercalative polymerization: In this technique, the layered silicate is swollen within the liquid monomer (or a monomer solution) so as the polymer formation can occur in between the intercalated sheets. Polymerization can be initiated either by heat or radiation, by the diffusion of a suitable initiator or by an organic initiator or catalyst fixed through cationic exchange inside the interlayer before the swelling step by the monomer.  Melt intercalation: The layered silicate is mixed with the polymer matrix in the molten state. Under these conditions and if the layer surfaces are sufficiently compatible with the chosen polymer, the polymer can crawl into the interlayer space and form either an intercalated or an exfoliated nanocomposite. In this technique, no solvent is required.  Template synthesis: This technique, where the silicates are formed in situ in an aqueous solution containing the polymer and the silicate building blocks has been widely used for the synthesis of double-layer hydroxide-based nanocomposites [14,15] but is far less developed for layered silicates. In this technique, based on self-assembly forces, the polymer aids the nucleation and growth of the inorganic host crystals and gets trapped within the layers as they grow. The following sections review the four aforementioned preparation techniques, that will be illustrated with representative examples. 3.1. Exfoliation±adsorption 3.1.1. Exfoliation±adsorption from polymers in solution This technique has been widely used with water-soluble polymers to produce intercalated nanocomposites [24,25] based on poly(vinyl alcohol) (PVOH) [26,27], poly(ethylene oxide) (PEO) [27±31], poly(vinylpyrrolidone) (PVPyr) [32] or poly(acrylic acid) (PAA) [31]. When polymeric aqueous solutions are added to dispersions of fully delaminated sodium layered silicates, the strong interaction existing between the hydrosoluble macromolecules and the silicate layers often trigger the reaggregation of the layers as it occurs for PVPyr [32] or PEO [27]. In the presence of PVOH, the layers remain in colloidal distribution [27]. In the wet state or after mild drying (air drying), the silicate layers are distributed and embedded in the so-obtained PVOH gel. This state actually corresponds to a true nanocomposite hybrid material. However, more intense drying of the PVOH gel in vaccuo causes part of the silicate layers to reaggregate and intercalated species are formed. This is indicated by a basal spacing of 1.36 nm, corresponding to the intercalation of a polymeric monolayer in between the silicate layers. In fact, sterical constraints from the PVOH matrix impede reaggregation of all the silicate layers and some of them remain exfoliated [33]. Interestingly, polymers intercalation using the so-called exfoliation±adsorption technique can also be performed in

7

8

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 7. X-ray diffractograms of modified montmorillonite, HDPE and nitrile copolymer composite systems (* Ð a high degree of crystallinity of the HDPE is evident) (reproduced from [35] with permission).

organic solvents. PEO has been successfully intercalated in sodium montmorillonite and sodium hectorite by dispersion in acetonitrile [34], allowing to stoichiometrically incorporate one or two polymer chains in between the silicate layers and increasing the intersheet spacing from 0.98 to 1.36 and 1.71 nm, respectively. Study of the chain conformation using two-dimensional double-quantum NMR on 13 C enriched PEO intercalated in sodium hectorite [10] reveals that the conformation of the `±OC±CO±' bonds of PEO is 905% gauche, inducing constraints on the chain conformation in the interlayer. Jeon and coworkers [35] have investigated this technique in attempts to produce nanocomposites with nitrile-based copolymer (Barex 210 E) and polyethylene-based polymer. These nanocomposites were filled with sodium montmorillonite previously modified by a protonated dodecylamine as the organic cation. Upon treatment, the interlayer spacing increased from 11.8 to Ê , attesting for effective cation exchange. In order to produce the nitrile copolymer-based 16.5 A nanocomposite, the copolymer was dissolved in dimethylformamide in the presence of 15 wt.% modified clay. After solvent evaporation in a vacuum oven at 808C for 24 h the film recovered was characterized by both XRD and TEM. XRD reveals a broad diffraction peak that has been shifted Ê , see Fig. 7). The large broadening of the peak may towards a higher interlayer spacing (21.5 A indicate that partial exfoliation has occurred, as corroborated by TEM analysis (Fig. 8) where both stacked (intercalated) and isolated (exfoliated) silicate layers can be observed. High density polyethylene (HDPE)-based nanocomposite has been produced by using a similar technique where the polyolefinic chains were dissolved in a mix of xylene and benzonitrile (80:20 wt.%) with 20 wt.% modified clay dispersed within. The composite material was then recovered by precipitation from tetrahydrofuran (THF) followed by several washings with THF. As seen in Fig. 7, the small increase in the interlayer spacing could account for some intercalation even if the TEM observation let only show small stacks of flake-like particles. Even if conducted under similar experimental conditions, these two syntheses indicate that the exfoliation±adsorption technique can provide quite different results much depending upon the polymer matrix. In other words, it does mean that for every type of polymer, one has to find the right layered clay, organic modifier and solvent(s). Ogata et al. applied the exfoliation±adsorption method for the production of poly(lactide) (PLA) [36] and poly(e-caprolactone) (PCL) biodegradable nanocomposites [37] using montmorillonite modified with distearyldimethylammonium cations. The composites were prepared by dissolving either PLA or PCL in hot chloroform in presence of a given amount of the modified clay, then

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 8. TEM micrograph of the nitrile copolymer filled with 15 wt.% of organo-modified montmorillonite. Iˆindividual silicate layer and Sˆstacked silicate layers (reproduced from [35] with permission).

vaporizing the solvent to obtain homogeneous films. However, under those conditions, it was found that no intercalation took place in the presence whatever polyester. It is worth to point out that the organo-modified clay rather formed a remarkable geometric structure in the filled polymers where tactoids consisting of several silicate monolayers form a superstructure in the thickness direction of the film. Such structural features have been found on one hand to substantially increase the Young's modulus of the PLA-based composites (which is almost doubled with 5 wt.% of organo-modified clay) and on the other hand, to enhance both storage and loss moduli determined by dynamic mechanical analysis (DMA) carried out on the organoclay-filled PCL. 3.1.2. Exfoliation±adsorption from prepolymers in solution Some polymeric materials such as poly(imides) or some conjugated polymers have the particular property of being infusible and insoluble in organic solvents. Therefore, the only possible route to produce nanocomposites with these types of polymers consists in using soluble polymeric precursors that can be intercalated in the layered silicate and then thermally or chemically converted in the desired polymer. This has been successfully achieved by using the exfoliation±adsorption process. The Toyota Research group has been the first to use this method to produce poly(imide) (PI) nanocomposites [38]. The polyimide±montmorillonite nanocomposite has been synthesized by mixing in dimethylacetamide a modified montmorillonite with the poly(imide) precursor, that is a poly(amic acid) obtained from the step polymerization of 4,40 -diaminodiphenyl ether with pyromellitic dianhydride. The organo-modified montmorillonite was prepared by previous intercalation with dodecylammonium hydrochloride. After elimination of the solvent, an organoclay filled poly(amic acid) film was recovered, which was thermally treated up to 3008C in order to trigger the imidization reaction and to produce the poly(imide) nanocomposite. The XRD patterns of these filled PI films do not show any diffraction peak typical of an intercalated morphology leading the authors to conclude to the formation of an exfoliated structure and explaining the excellent gas

9

10

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 3 Nature, CEC and length of the layered silicates used in the synthesis of polyimide as presented in [39] Layered silicate ‡

Hectorite Na Saponite Na‡ Montmorillonite Na‡ Synthetic mica Na‡ a

CEC (meq/100 g)

Ê )a Length of dispersed particles (A

55 100 119 119

460 1650 2180 12300

Longer particle dimension as determined by TEM observation.

barrier properties of the resulting films (see Section 4.3). This experiment has been extended to other layered silicates (hectorite, saponite and synthetic mica) with different aspect ratios [39] (Table 3). X-ray diffractograms of the obtained polyimide-based nanocomposites again show no noticeable peak indicating an exfoliated structure for the montmorillonite and the synthetic mica. Ê is observed, indicating that For both hectorite and saponite, a broad peak, centered on a value of 15 A for those layered silicates, polymer intercalation occurs probably together with some exfoliation. For saponite, the measured interlayer spacing is even smaller than the value measured for the starting Ê ) suggesting that the organic cation could have been expelled from organically modified clay (18 A the clay interlayer during imidization reaction. The same phenomenon has been observed by Lan et al. [40] when studying the effect of the chain length of the organic cation in the preparation of PI nanocomposites by the same synthetic methodology. Starting from sodium montmorillonite with a CEC of 92 meq/100 g and various protonated linear primary alkylamine, i.e. CH3±(CH2)nÿ1NH3‡ (where nˆ4, 8, 12, 16 and 18), they obtained, after imidization by curing at 3008C, composites Ê , independently of the chain length of the intercalated showing the same interlayer spacing of 13.2 A alkylammonium cation (Fig. 9).

Fig. 9. XRD patterns of polymer/CH3(CH2)nÿ1NH3‡ modified montmorillonite composites (clay loading: 10 wt.%): (A) air-dried poly(amic acid) films and (B) poly(imide) films cured at 3008C (reproduced from [40] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

This value again corresponds to the intercalation of the PI chains adopting a flattened conformation within the galleries, and the removal of the organic cation out of the interlayer. As the same is achieved at much lower temperature (drying at 1008C), this structural change cannot be related to the thermal degradation of the onium ion still stable in this temperature range. Further experimental evidence for the cation eviction out of the clay interlayers comes from the fact that the Ê interlayer spacing does not change when the material is heated up to at 4508C, a temperature 13.2 A where alkylammonium cations are usually degraded. No clear explanation for the expulsion of the alkylammonium ions has been reached. The presence of organically modified layered silicate such as montmorillonite previously modified with protonated p-phenylene diamine has also shown to improve the kinetics of the imidization reaction, allowing for a reduction of both the imidization temperature and reaction time [41]. The activation energy of the imidization reaction (based on a first order kinetics), monitored by FTIR spectroscopy, is shown to drop down by ca. 20% in presence of 7 wt.% of organoclay. A reaction mechanism has been tentatively proposed, involving the modified silicate layers as active partners in the imidization process (Fig. 10). It has to be noted that a complete exfoliated structure has been observed for those nanomaterials by both XRD and TEM. Conjugated polymers are another family of polymers prone to be intercalated through this twostep technique. Oriakhi et al. [42] have elegantly shown that the exfoliation±adsorption method could be explored to prepare nanocomposites with poly(p-phenylenevinylene) (PPV) as the continuous polymeric matrix. The polymer precursor to be intercalated was the poly(p-xylenylene dimethylsulfonium bromide) (PXDMS). The PXMDS-montmorillonite layered nanocomposite was accordingly prepared by reaction of a colloidal dispersion of Na-montmorillonite with an aqueous solution of PXMDS at 08C. As the precursor bears two cationic sites, it readily intercalates between the montmorillonite sheets by cationic exchange. The precursor is then chemically transformed into PPV by a base-mediated elimination of dimethylsulfide and HBr. This is achieved by stirring the crude product with 20% ethanolic NaOH solution at ambient temperature for 48 h. XRD patterns Ê , respectively. before and after chemical conversion give interlayer spacing of 15.1 and 14.6 A Ê ), it indicates a gallery Compared to the initial interlayer spacing of the Na-montmorillonite (9.6 A Ê expansion of, respectively, 5.5 and 5.0 A, consistent with the expected dimensions for a polymer

Fig. 10. A possible reaction mechanism for involving silicate layers in the imidization process (reproduced from [41] with permission).

11

12

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

monolayer with the phenyl rings oriented perpendicular to the layered silicate surfaces. This gallery expansion together with the absence of any XRD peak related to the interlayer spacing of Namontmorillonite attest for the formation of an intercalated nanocomposite. The presence of a weak absorption bandpeak in the sp3 C±H region of the FTIR spectrum (2890 cmÿ1), however, suggests that the elimination of the dimethylsulfonium groups is not quantitative and that the incorporated polymer could be in fact a copolymer of PXMDS and PPV. 3.1.3. Exfoliation±adsorption by emulsion polymerization Emulsion polymerization has been also studied in order to promote the intercalation of water insoluble polymers within Na-montmorillonite that is well known to readily delaminate in water [43±45]. Poly(methyl methacrylate) (PMMA) was first tested by this method [43]. The emulsion polymerization was thus carried out in water in the presence of various amounts of the layered silicate. The previously distilled methyl methacrylate monomer (MMA) was dispersed in the aqueous phase with the aid of sodium lauryl sulfate as a surfactant. Polymerization was conducted at 708C for 12 h by using potassium persulfate as the free-radical initiator. The obtained latex is then coagulated with an aluminum sulfate solution, filtered and dried under reduced pressure. The obtained composites were extracted with hot toluene for 5 days by means of Soxhlet extraction. Contents of intercalated polymer were determined for both extracted and non-extracted materials and are given in Table 4 together with the molecular weights and polydispersities of the extracted polymers. These results demonstrate that part of the PMMA chains stay immobilized onto and/or inside the layered silicates and cannot be extracted. This is further confirmed by FTIR of the extracted composite that shows the absorption bands typical of PMMA chains. It can be observed that the relative content of clay does not substantially modify the PMMA molecular weights (Mw), the value of which is quite comparable to the Mw of PMMA polymerized in absence of clay (entry 1, Table 4). Clearly, the presence of layered silicates does not seem to perturb the free-radical polymerization. Ê is Intercalation is evidenced by XRD where an increase in the interlayer distance of about 5.5 A observed for both PMMA 10, 20 and 30. This increase relatively well correlates with the thickness of the polymer chain in its extended form. DSC data obtained for the extracted nanocomposites does not show any glass transition, in accordance with what is usually observed for intercalated polymers. Ion±dipole interactions are believed to be the driving force for the immobilization of the organic polymer chains lying flat onto the layered silicate surface. The same methodology has been also applied to produce montmorillonite intercalated with poly(styrene) (PS) [45]. The nanocomposite Table 4 Montmorillonite feed ratios, PMMA contents in non-extracted and extracted composites, average molecular weights and polydispersities of extracted PMMAs Sample PMMA PMMA10 PMMA20 PMMA30 PMMA40 PMMA50 a b

Feed ratio of MMA/clay (g/g)

PMMA content (wt.%)a Non-extracted

Extracted

100/0 100/10 100/20 100/30 100/40 100/50

± 87.4 79.3 60.4 58.6 46.1

± 58.7 49.6 33.4 22.8 18.4

b

As determined by TGA. Composite recovered after Soxhlet extraction in toluene for 5 days.

Mn10ÿ3 (g/mol)

Mw10ÿ3 (g/mol)

Mw/Mn

23 44 60 63 82 38

160 250 200 150 390 290

6.6 5.8 3.4 2.4 4.8 7.6

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 5 Montmorillonite feed ratios, PS contents in non-extracted and extracted composites and interlayer distances in PS-based nanocomposites obtained by emulsion polymerization Sample MMT PS5 PS10 PS20 PS30 a b

Feed ratio styrene/MMT (g/g) 0/100 95/5 90/10 80/20 70/30

Ê) Interlayer distance (A

PS content (wt.%)a Non-extracted

Extracted

± 88.7 82.8 75.4 65.9

± 45.6 33.9 28.8 21.7

b

9.8 15.5 14.6 13.8 12.4

As determined by TGA. Composite recovered after Soxhlet extraction in toluene for 5 days.

syntheses were essentially comparable to MMA emulsion polymerization except that the Namontmorillonite was sonicated prior to polymerization. Results are presented in Table 5. Here again, a true intercalated structure is formed with an interlayer spacing that changes with the PS content. The interlayer distance slightly decreases at higher montmorillonite feed ratios. Similarly to MMA polymerization, the molecular weight of the PS recovered fraction does not seem to be affected by the presence of the dispersed clay. DSC thermograms also attest for the intercalation of PS in an extended form as no more glass transition can be observed in the extracted nanocomposites. However, in the case of filled PS, the stabilization of the PS chains in the interlayer cannot be accounted for by ion±dipole interactions anymore. Rather, the authors propose the cooperative formation of ion-induced dipole interactions. Note finally a report on the formation of epoxy-montmorillonite intercalated composites by emulsion polymerization [44]. Again an increase Ê is observed. But contrary to observations achieved for both of the interlayer spacing of about 6 A PMMA and PS, the epoxy content in the extracted composites appears to increase with the montmorillonite content, indicating that the layered silicates possibly participate in the polymerization reaction. 3.2. In situ intercalative polymerization 3.2.1. Thermoplastic nanocomposites Many interlamellar polymerization reactions were studied in the 1960s and the 1970s using layered silicates (see [25,46] and references therein) but it is with the work initiated by the Toyota research team [47,48] that the study of polymer-layered silicate nanocomposites came into vogue about 10 years ago. They studied the ability of Na-montmorillonite organically modified by protonated a,o-aminoacid (‡H3N±(CH2)nÿ1±COOH, with nˆ2, 3, 4, 5, 6, 8, 11, 12, 18) to be swollen by the e-caprolactam monomer (melting temperatureˆ708C) at 1008C and subsequently to initiate its ring opening polymerization to obtain nylon-6-based nanocomposites [48,49]. A clear difference occurs in the swelling behavior between the montmorillonite with relatively short (n<11) and longer alkyl chains as depicted in Table 6, indicating that a larger amount of monomer can be intercalated for longer alkyl chains. The 12-aminolauric acid (nˆ12) modified montmorillonite was chosen to develop the intercalative ring opening polymerization of e-caprolactam [48]. In a typical synthesis, the modified montmorillonite (12-Mont) was mixed with the monomer in a mortar. A small amount of 6aminocaproic acid was added as a polymerization accelerator when the relative amount of 12-Mont used was smaller than 8 wt.% (relative to 12-Mont). The mixture was heated first at 1008C for

13

14

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 6 Basal spacing on organo-modified montmorillonite in the presence of e-caprolactam (e-CLa) at 1008C ‡

H3N±(CH2)nÿ1±COOH (n)

2 3 4 5 6 8 11 12 18

Interlayer spacing of the Ê) modified clay (A

Interlayer spacing when swollen Ê) by (e-CLa) at 1008C (A

12.7 13.1 13.2 13.2 13.2 13.4 17.4 17.2 28.2

14.4 19.7 19.9 20.4 23.4 26.4 35.7 38.7 71.2

30 min then at 2508C for 6 h. The cooled and solidified product was crushed, washed with water at 808C, and then dried. Depending on the amount of 12-Mont introduced, either exfoliated (for less then 15 wt.%) or delaminated structure (from 15 to 70 wt.%) were obtained as evidenced by XRD and TEM measurements. Comparison of the titrated amount of COOH and NH2 end groups present in the synthesized nanocomposites with given values such as the CEC of the montmorillonite used (119 meq/100 g) have led to the conclusion that the COOH end groups present along the 12-Mont surface are responsible for the polymerization initiation. Moreover, approximately all the `‡ NH3 ' end groups present in the matrix should be interacting with the montmorillonite anions. Finally, the ratio of bonded to non-bonded polymer chains increased with the amount of incorporated montmorillonite (from 32.3% of bonded chains for 1.5 wt.% montmorillonite to 92.3% of bonded chains for 59.6 wt.% clay). Further works [50] have demonstrated that intercalative polymerization of e-caprolactam could be realized without the necessity to render the montmorillonite surface organophilic. Indeed, this monomer was able to directly intercalate the Na-montmorillonite in water, in the presence of hydrochloric acid. This intercalation was proved by the increase in interlayer Ê for the spacing observed on the isolated montmorillonite/e-caprolactam product, going from 10 A Ê when e-caprolactam intercalates. At high temperature (2008C) in neat Na-montmorillonite to 15.1 A the presence of excess e-caprolactam, the so modified clay can be swollen again, allowing for the ring opening polymerization to proceed at 2608C when 6-aminocaproic acid is added as an accelerator. The resulting composite does not present any diffraction peak characteristic of an interlayer spacing in XRD and TEM observation agrees with a molecular dispersion of the silicate sheets. In attempts to carry out the synthesis in one-pot [51], the system has proved to be sensitive to the nature of the acid used to promote the intercalation of e-caprolactam. Table 7 gives the results Table 7 Peak intensity (Im) and interlayer spacing (d) of nylon-6-based nanocomposites prepared in presence of different acid derivatives by the one-pot technique Acid

Ima (cps)

Ê) d (A

Phosphoric acid Hydrochloric acid Isophtalic acid Benzenesulfonic acid Acetic acid Trichloroacetic acid No acid

0 200 255 280 555 585 1840

0 21.7 20.2 19.3 20.3 21.3 18.6

a

As defined in Fig. 11.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 11. XRD intensity curve of injection molded nylon-6 nanocomposite as obtained by the one-pot intercalation polymerization process in the presence of acetic acid (reproduced from [51] with permission).

obtained for different acids in relation to the intensity (Im) of the XRD intercalation peak that might be present in the obtained nanocomposites (Fig. 11). These results show that only phosphoric acid allows for the preparation of a truly exfoliated nanocomposite, the other acids tending to promote the formation of partially exfoliated-partially intercalated structures. No clear reason indicates why only phosphoric acid works. One can also point out that an intercalated structure is obtained even if no acid is added on purpose. Actually, the addition of 6-aminocaproic acid in each experiment as polymerization accelerator could have played the same role. Another polyamide, i.e. nylon-12, is reported to form nanocomposites using the in situ intercalative polymerization. Indeed, Reichert et al. [52] have used 12-aminolauric acid (ALA) as both the layered silicate modifier and the monomer. They first studied by XRD the effect of ALA on the swelling behavior of a synthetic three-layer silicate (commercial SOMASIF ME100, a fluorinated silicate obtained by heating talcum and Na2SiF6 at high temperature for several hours). Increasing amounts of ALA dispersed in a constant HCl volume (20 mmol/l) were poured into a water suspension of ME100. The swelling process in function of ALA concentration was monitored and it can be separated in two regimes: a cation-exchange of inorganic cations by protonated ALA at Ê and a further diffusion of zwitterionic low ALA concentration giving an interlayer distance of 17 A 12-aminolauric acid into the interlayer spacing when the ALA concentration exceeds the amount of Ê , Figs. 12 and 13). HCl in the medium (interlayer spacing over 20 A The swelling was found to be independent of both the swelling temperature, the layered silicate concentration and the type of mineral acid used to protonate ALA (HCl, H2SO4, H3PO4). ALA was

Fig. 12. Interlayer distance of fluoro-modified talc (ME 100) in function of an increasing amount of aminolauric acid used as the organic modifier (reproduced from [52] with permission).

15

16

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 13. Schematic representation of the swelling behavior of the fluoro-modified talc ME 100 in presence of aminolauric acid (reproduced from [52] with permission).

then polymerized at high temperature (2808C) and under elevated pressure (up to 20 bar) during 9.5 h with both types of swollen clay. XRD, TEM, Scanning TEM coupled with energy dispersive X-ray (EDX) as well as atomic force microscopy (AFM) have been used to characterize the resulting composites. They all confirm the structure as being partially exfoliated and otherwise intercalated nanocomposites. By using the method developed by Usuki et al. [48] for the polymerization of e-caprolactam (vide supra), Messersmith and Giannelis [53] have modified a Na-montmorillonite by the protonated aminolauric acid and dispersed this modified clay in liquid e-caprolactone before polymerizing it at high temperature. The poly(e-caprolactone) (PCL) nanocomposites were prepared by mixing up to 30 wt.% of the modified clay with dried and freshly distilled e-caprolactone at room temperature for a couple of hours followed by the ring opening polymerization under stirring at 1708C for 48 h. XRD patterns of the modified clay after contact with e-caprolactone at room temperature do not show any Ê ). The authors assumed that the monomer significant increase in the layered spacing (13.6 A intercalates in the gaps between the aminolauric acid chains so that no gallery expansion could be seen. This is in contrast with what is usually observed in in situ intercalative polymerization where the insertion of the monomer within the silicate gallery induces an increase in the interlayer spacing. Another possibility may be that intercalation of the monomer occurs only during the heating step of the solution. After polymerization, XRD patterns of the obtained composites did not show any diffraction peak indicating the presence of intercalated structures, more likely indicating that exfoliation occurred. This exfoliation induces some clear modifications in the onset of polymer melting (Tm) of the obtained nanocomposites that appears at lower temperature when the weight fraction of layered silicates increases (Fig. 14). This phenomenon is attributed to the formation of smaller crystallites, probably due to the physical barrier effect of the dispersed layers that limits the extent of crystallization of the PCL chains. The authors propose a polymerization mechanism where the carboxylic acid of the attached aminolauric acid undergoes a nucleophilic addition on the e-caprolactone carbonyl function, creating by ring opening reaction a carboxylic anhydride bridge that links the first opened lactone monomer to the silicate surface. The ring opening reaction is then proposed to occur via an `o-acyl' cleavage of the monomer with the formation of an hydroxyl end group that propagates the lactone polymerization. However, it is now commonly accepted [54] that carboxylic species are not reactive enough to promote the ring-opening polymerization of e-caprolactone. At such a high temperature (1708C), a mechanism known as the `active hydrogen polymerization' should rather take place where traces of water could initiate the ring opening polymerization of the lactone [55]. This

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 14. Polymer onset melting temperature as a function of organomodified montmorillonite for poly(e-caprolactone) (PCL)-based nanocomposites. PCLA corresponds to the melting temperature for unfilled PCL (reproduced from [53] with permission).

mechanism ruled out the grafting of the growing polyester chains onto the silicate layers through the formation of a carboxylic anhydride link. If any links exist (as suggested by indirect evidence such as the broadening of the solution 1 H-NMR spectra of the composites when the amount of layered silicates increases or by the possibility to recover the PCL chains free of silicates after cation exchange with Li‡), they would rather be formed by a post-esterification reaction at high temperature between the terminal hydroxyl group of some PCL chains with the carboxylic acid function of the fixed aminoacid surface agent. Note finally that e-caprolactone ring opening polymerization could be also initiated by primary amino end-groups of some aminolauric acids that did not react with the layered silicate surface. Indeed, Rozenberg has claimed that e-caprolactone polymerization initiated with amines in the presence of a protic acid at a high temperature (1808C) can proceed through a zwitterionic mechanism [56]. The resulting macromolecular zwitterions, i.e. a-H3N‡, o-CO2ÿ PCL chains could then interact with the charged layered silicates, assuring some efficient surface grafting. The above poly(e-caprolactone)-based nanocomposite synthesis has been recently applied by Chen et al. [57] to produce novel segmented polyurethane/clay nanocomposites articulated on diphenylmethane diisocyanate, butanediol and preformed polycaprolactone diol. Even if the mechanism proposed for the chemical link between the nanofiller surface and the polymer does not appear appropriate (ammonium salts are not known to induce e-caprolactone ring-opening polymerization), they succeeded in producing a material where the nanofiller acts as a multifunctional chain extender inducing the formation of star-shaped segmented poly(urethane). Messersmith and Giannelis [58] have also reported on the e-caprolactone polymerization inside a Cr3‡-exchanged fluorohectorite at 1008C during 48 h. Interestingly enough, the monomer Ê intercalation has been this time evidenced by the interlayer spacing increase from 12.8 to 14.6 A upon the addition of the liquid monomer. After the polymerization reaction, an intercalated poly(eÊ observed by XRD, due to the caprolactone) is obtained as proved by an interlayer spacing of 13.7 A dimensional change accompanying the polymerization of the cyclic monomer as sketched in Fig. 15. Ê correlates well with the sum of the thickness of the The observed layer spacing of 13.7 A Ê Ê ) in the crystal structure of poly(esilicate layer (9.6 A) and the known interchain distance (4.0 A

17

18

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 15. XRD patterns of e-caprolactone intercalated in Cr3‡ modified fluorohectorite (plain line) and the resulting poly(ecaprolactone)-based nanocomposite (dashed line). Insets are schematic illustrations corresponding the intercalated monomer (left) and intercalated polymer (right) (reproduced from [58] with permission).

caprolactone). Extraction of unintercalated polymer with acetone allows to analyze the intercalated composite that contains 7.7 wt.% of the polymer. DSC analysis of this extracted nanocomposite does not show any melting endotherm in the vicinity of 608C, that is the expected melting temperature of the PCL. This effect is attributed to the polymer chains confinement within the silicate galleries that prevents the formation of polymer crystallites. In situ intercalative polymerization has also been largely studied for producing poly(styrene)based nanocomposites. Akelah and Moet [59] modified the interlayer of Na-montmorillonite and Camontmorillonite by exchanging the inorganic cations with (vinylbenzyl)trimethyl ammonium Ê . These modified clays were then dispersed and chloride, increasing the interlayer spacing by 5.4 A swollen in various solvent and cosolvent mixtures such as acetonitrile, acetonitrile/toluene and acetonitrile/THF. Styrene polymerizations were carried out in presence of N,N0 -azobis(isobutyronitrile) (AIBN) and carried out at 808C for 5 h. The composites were isolated by precipitation of the colloidal suspension in methanol, filtered off and dried. By this technique, intercalated composites Ê depending on the nature of were produced with interlayer spacings varying between 17.2 and 24.5 A the solvent used. Even if the polymer is well intercalated, a drawback of this technique remains that the macromolecule produced is not a pure PS but rather a copolymer between styrene and (vinylbenzyl)trimethyl ammonium cations. Do and Cho have developed a technique using more commonly modified montmorillonites [60]. The authors compared the ability of several tetraalkylammonium cations incorporated into Namontmorillonite through ion-exchange to promote the intercalation of poly(styrene) through the free radical polymerization of styrene initiated by AIBN at 508C. Three tetraalkylammonium cations were tested, all based on the following formula: (CH3)2N‡(hydrogenated tallow alkyl)R where hydrogenated tallow alkyl corresponds to a mixture of mainly octadecyl chains together with small amounts of lower linear homologues and R may be either another hydrogenated tallow alkyl (Ta), 2ethyl hexyl (Eh) or benzyl (Bz) group. These so-modified organo-montmorillonites were coded as Ta-MMT, Eh-MMT and Bz-MMT, respectively. Layer spacings obtained for the three MMTs and corresponding composites are presented in Table 8. An evaluation of the filler dispersibility within styrene during the polymerization reaction is provided as well.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

19

Table 8 Interlayer spacings of organo-modified montmorillonites (X-MMT) and as obtained PS-based nanocomposites and the clay dispersibility within the polymerization medium Xa-MMT Sodium MMT Bz-MMT Eh-MMT Ta-MMT

Ê) Interlayer spacing (A

Dispersibilityb

In X-MMT

In PS/X-MMT

11.8 19.1 20.4 32.7

14.2 34.0 28.5 32.9

± ‡‡ ‡ ‡

a

Organo-modifiers: (CH3)2N‡(hydrogenated tallow alkyl)R with RˆBz (benzyl), Eh (2-ethylhexyl), or Ta (hydrogenated tallow alkyl). b It was judged by the appearance of the montmorillonite dispersion in styrene monomer: (‡‡) fully dispersible; (‡) partly dispersible; (±) non-dispersible.

It turns out that the best intercalation occurs for Bz-MMT. This is probably due to a better affinity between styrene and benzyl groups spread all along the layered montmorillonite surfaces as further demonstrated by the perfect dispersibility of this organo-modified filler in styrene. If EhMMT does also intercalate some PS, the interlayer spacing for Ta-MMT does not change a lot (only Ê increase) but TGA analysis seems to indicate that intercalation does occur in this composite a 2A too (see Section 4.2.1 for further information). According to authors' discussion, intercalation within Ta-MMT should occur without an important increase in the interlayer spacing because such hydrogenated tallow alkyl chains should be long enough (mainly C18 chains) to easily accommodate PS. Even though this technique allows an extensive intercalation of PS chains through an adequate choice of the alkylammonium cation neither exfoliation nor control over the molecular parameter of the polymer (PS) produced have been observed. Such a control has been, however, achieved by Weimer et al. [61] who modified a Namontmorillonite by anchoring an ammonium cation bearing a nitroxide moiety known for its ability to mediate the controlled/`living' free radical polymerization of styrene. The intercalative polymerization principle used is sketched in Fig. 16. Styrene polymerization was carried out in bulk (in absence of solvent) at 1258C for 8 h yielding homogeneous and transparent composites. The absence of diffraction peaks in the low angle area of the XRD patterns together with the TEM observations of silicate layers randomly dispersed within the PS matrix attest for the complete exfoliation of the layered silicate. The polymer chains were desorbed from the silicate layers by refluxing the nanocomposite in THF in the presence of LiCl. Remarkably, the PS number average molecular weight measured by size exclusion chromatography (SEC) (Mnˆ21,500) was in perfect agreement with the theoretical Mn expected from the initial monomer to nitroxide initiator molar ratio, assuming a living polymerization (Mn theoreticalˆ24,400). Furthermore, the molecular weight distribution was monomodal with a narrow polydispersity of 1.3. It is worth pointing out that the tailoring of the PS molecular weight can be achieved by varying the amount of fixed initiator. For doing so, variable quantities of an inactive alkylammonium (trimethylbenzylammonium) was added together with the nitroxide-bearing ammonium, during the ion-exchange reaction. Thus, this strategy allows to tune up with the molecular weight of the recovered PS while keeping a narrow polydispersity and an exfoliated structure. `Livingness' of the free-radical polymerization was definitively evidenced by a quantitative resumption and chain extension experiment. Indeed, addition of a second styrene feed to the reaction medium led to a Mn increase from 30,000 to 74,000 after complete monomer consumption, well in agreement with the calculated value of 75,300. It has to be noted that this

20

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 16. Schematic representation of the montmorillonite modification with nitroxyl-based organic cation and its subsequent use to produce PS-based exfoliated nanocomposite (reproduced from [61] with permission).

technique is the only known way to obtain complete layered silicate exfoliation in a PS matrix, still unreachable trough exfoliation±adsorption (see Section 3.1) or melt intercalation processes (see Section 3.3). Polyolefins represent another important family of polymers that has been investigated in order to produce nanocomposites through in situ intercalative polymerization process. Tudor et al. [62] have demonstrated the ability of soluble metallocene catalysts to intercalate inside silicate layers and to promote the coordination polymerization of propylene. Accordingly, a synthetic hectorite (Laponite RD) was first treated with methylaluminoxane (MAO) in order to remove all the acidic protons and to prepare the interlayer spacing to receive the transition metal catalyst. It has to be

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

noted that MAO is commonly used in association with metallocenes to produce coordination catalysts active in olefin polymerization. During this first treatment step, no noticeable increase of the layer spacing was observed (even if the diffraction peak broadened slightly) but both increase in Al content and IR data showing complete disappearance of absorptions assignable to Si±OH groups agree with the MAO reaction/adsorption inside the layered silicate galleries. Upon the addition of the metallocene catalyst ([Zr(Z-C5H5)2Me(THF)]‡), a cation exchange occurs with Na‡ and the catalyst is incorporated in the hectorite galleries as demonstrated by an increase in the interlayer spacing of Ê , consistent with the size of the catalyst species. Using a synthetic fluorinated mica-type silicate 4.7 A that is deprived from any protons in the galleries, the catalyst was even incorporated directly within the filler interlayer, without the need of MAO pretreatment. These two modified layered silicates catalyzed with reasonably high activity the polymerization of propylene when contacted with an excess of MAO, producing PP oligomers. Unfortunately, no composite characterization was provided with, so as one cannot claim intercalated or exfoliated morphology. In a very recent work [63,64], intercalated/exfoliated nanocomposites based on high density polyethylene matrices have been synthesized by the so-called polymerization-filling technique (PFT) [65]. This method consists in anchoring in a first step, a Ziegler±Natta type catalyst or any other coordination catalysts, which include MAO activated metallocenes, onto a filler surface, and then in situ polymerizing ethylene and/or a-olefins directly from the surface treated fillers. High performance microcomposites, combining both high stiffness and toughness, have been produced by PFT, as a result of homogeneous filler dispersion, strong filler/matrix interfacial adhesion and the unique possibility to get highly filled ultrahigh molecular weight polyethylene. It was, therefore, of prime interest to apply this technique to nanofillers. Layered silicates (montmorillonite and hectorite) in aqueous clay colloidal suspension were made less hydrophilic through the elimination of water by freeze drying. The obtained fluffy materials could then be nicely dispersed in a non-polar solvent such as heptane or toluene. The clay dispersion was then surface treated with MAO and, after solvent removal by evaporation, a high temperature treatment at 1508C was applied to modify the layered silicate. After washing the modified clay with toluene in order to remove unreacted MAO, the silicate layers were contacted with a metallocene precatalyst, i.e. (tert-butylamido) dimethyl (tetramethyl-Z5-cyclopentadienyl) silane titanium dimethyl and aged for 1 h in order to form the active catalyst species. The polymerization is then carried out by adding ethylene in the medium. It has to be pointed out that contrary to the procedure reported by Tudor et al. (vide supra), no ion exchange reaction was required since the added metallocene precatalyst was not a cationic one. Rather, the strategy here relies upon the immobilization of the active species through electrostatic interactions with surface anchored MAO as already performed with more conventional microfillers (kaolin, silica) [66]. Some typical in situ intercalative polymerization experiments are gathered in Table 9. Polymerizations were conducted in highly diluted conditions (ca. 2 g filler per liter of dried heptane) so that a low level of layered silicates was reached in the final composition (down to 3 wt.%). When ethylene polymerization was carried out in the absence of molecular hydrogen (thus without any transfer agent), layered silicate filled UHMWPE is produced which is an extremely viscous material and is highly difficult to melt (Table 9, entries 1 and 2). Addition of hydrogen to the polymerization medium allows molecular weight to be reduced down (Mwˆ504,000 and Mnˆ77,000 as determined by SEC) with substantial improvement of the melt processability (Table 9, entry 3). Examination of the TEM pictures reveals that layered silicate exfoliation partially occurred. For instance, at high filler content (ca. 30 wt.% montmorillonite), besides some stacked multilayered silicates, one can observe well exfoliated and dispersed nanosized sheets (Fig. 17).

21

22

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 9 Synthesis and composition of PE-based nanocomposite produced by in situ intercalative polymerization of ethylene (P(C2H4)ˆ10 bar) in non-organo-modified layered silicatesa Filler

MAO (10ÿ3 mol)

Catalyst (10ÿ6 mol)

P(H2)b (bar)

Filler loadingc (wt.%)

HPDE Mn (g/mol)

h m h

33.00 27.20 23.75

15.6 12.5 16.2

0 0 0.3

4.2 3.3 3.4

±d ±d 77,000

a

hˆhectorite and mˆmontmorillonite. Hydrogen partial pressure at start. c Measured by thermogravimetric analysis (TGA). d Insoluble UHMWPE that cannot be eluted by SEC. b

Heinemann et al. [67] have also investigated intercalative polymerization to produce (co)polyolefin nanocomposites. They carried out the polymerizations in the presence of modified layered silicates such as bentonite after sodium exchange with dimethyldistearylammonium or dimethylbenzylstearylammonium cations. For the sake of comparison, non-organo-modified clays were studied as well, including both synthetic hectorite and fluoromica. As a typical synthesis, the filler silicate was dispersed in toluene (or methylene chloride) followed (when needed) by the

Fig. 17. TEM micrograph of Na-montmorillonite exfoliated in HDPE after in situ intercalative polymerization of ethylene.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

addition of 1-octene as a comonomer. The reactor was then saturated with ethylene and polymerization was started by the addition of the catalyst. Ethylene was continuously introduced during polymerization. After a given time, the polymerization was stopped by addition of 2-propanol and the composite recovered by precipitation from acidified methanol, then filtered and dried. Different catalysts were used, affording HDPE and ethylene±octene copolymers (zirconium-based catalyst), and branched polyethylene (nickel and palladium-based catalysts). It appeared that the modified bentonites had a dramatic negative effect on the polymerization activity of the zirconiumbased catalyst, while Ni- and Pd-based catalysts were much less affected by the nature of the clay. This effect was attributed to the high sensitivity of the Zr-based active species towards any kind of polar functionality, including the anionic silicate layers covered by the alkylammonium cations. Nanocomposites were observed to be formed when polymerization was carried out by using organomodified clays. At the opposite, composites prepared either by in situ polymerization with nonmodified silicates or by melt-blending a preformed HDPE with a modified bentonite, only gave microcomposites as demonstrated by TEM and XRD. The presence of n-alkyl branches along the polyethylene chains (as a result of 1-octene copolymerization or migratory insertion ethylene polymerization promoted by Ni and Pd-based catalysts) was reported to enhance compatibility between the (co)polyolefin matrix and the dispersed layered silicates, improving the mechanical properties of the resulting nanocomposite materials. A very recent report discusses also about the synthesis of poly(ethylene terephtalate) (PET) nanocomposites by using the in situ intercalative polymerization [68]. The montmorillonite, Ê ), is reported modified by organic onium not given by the authors (interlayer distance: 11.2 A to react with PET comonomers (ethylene glycol and terephtalic acid derivatives) to form an Ê depending on the clay intercalated nanocomposite with an interlayer distance going from 14 to 35 A content. 3.2.2. Thermoset nanocomposites Next to all the aforementioned thermoplastic nanocomposites, in situ intercalative polymerization has also been explored to create thermoset-based nanocomposites. This method has been widely described for the production of both intercalated and exfoliated epoxy-based nanocomposites. Messersmith and Giannelis [69] have analyzed the effect of different curing agents and curing conditions in the formation of nanocomposites based on the diglycidyl ether of bisphenol A (DGEBA) and a montmorillonite modified by bis(2-hydroxyethyl)methyl hydrogenated tallow alkyl ammonium cation. They found that this modified clay dispersed readily in DGEBA when sonicated for a short time period as determined by the increase in viscosity at relatively low shear rates and the transition of the suspension from opaque to semitransparent. The increase in viscosity was attributed to the formation of a so-called `house-of-cards' structure in which edge-to-edge and edge-to-face interactions between dispersed layers form percolation structures. XRD patterns of uncured clay/ DGEBA also indicate that intercalation occurred during mixing and that this intercalation improves going from room temperature to 908C (Fig. 18). At room temperature a mix of intercalated and unintercalated clay species coexists as demonstrated by the persistence of the shoulder in the diffraction peak at 2yˆ5.88 and the appearance of the intercalation peak at 2yˆ2.58. At increasing temperature, the diffraction peak Ê keeping a constant intensity, suggesting that little or no delamination slightly shifts from 36 to 38 A occurs at or below 1508C. The influence of the curing agent used in the polymerization process appeared to be determinant in the resulting structure of the formed nanocomposite. When diamines were used, only intercalated epoxy-clay structures could be obtained. One possibility for this behavior may arise from the bridging of the silicate layers by the bifunctional amine molecules,

23

24

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 18. XRD patterns of organomodified montmorillonite, its uncured mixture with epoxy precursor (DGEBA) at room temperature and after 1 h annealing at 908C (reproduced from [69] with permission).

which prevents further expansion of the layers from taking place. When other curing agents such as nadic methyl anhydride (NMA), boron trifluoride monomethylamine (BTFA) or benzyldimethylamine (BDMA) were added, delamination during heating of the reaction mixture occurred as attested by XRD for BDMA (Fig. 19). Ê, Addition of the curing agent induced first an increase of the interlayer from 36 to 39 A indicating some partial intercalation. With further heating, disappearance of the interlayer spacing reflection indicated that delamination occurred. XRD of completely cured nanocomposites showed no more evidence of ordered reflections and TEM analysis displays silicate platelets separated from Ê . Moreover, study of the curing reactions tended to prove that the the others by more than 100 A particular alkylammonium used (that bears two hydroxyl functions) could play an active role, especially when BDMA or NMA were added as the curing agents. Indeed, BDMA can catalyze the

Fig. 19. XRD patterns of organo-modified montmorillonite/DGEBA/BDMA mixture (4 wt.% filler) heated in situ to various temperatures. The spectra are displaced vertically for clarity, with scan temperatures increasing from bottom to top (reproduced from [69] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

reaction between the hydroxyl groups of the alkylammonium and the oxirane of the monomer, producing a new hydroxyl that subsequently reacts with free DGEBA via a similar base-catalyzed oxirane ring opening to build up the epoxy network. As far as NMA is concerned, the resin formation may occur following a series of reactions beginning with the opening of the cyclic anhydride by the hydroxyl group of the alkylammonium. At higher temperature, the formed carboxylic acid can add onto the oxirane, liberating an hydroxyl group that reacts further with the oxirane groups to form the epoxy network. This mechanism is accredited by the fact that a complete curing cannot occur by heating DGEBA and NMA only. Furthermore, the DSC analysis shows two exotherms at 180 and 2478C that could correspond to the above two-step reaction. Finally, one can note the synthesis of nanocomposites based on montmorillonite and unsaturated polyester [70]. In this study, the montmorillonite was treated with a methacrylate±silane coupling agent in order to render the filler hydrophobic and reactive. The unsaturated polyester was polymerized by free radical polymerization with the modified montmorillonite dispersed in it. The authors claim the formation of exfoliated structure confirmed by the absence of diffraction peaks in the XRD pattern and a TEM micrograph showing a molecular dispersion of the clay layers. This example is the first one presenting an alternative way of modifying layered silicates by silane coupling reaction rather than interlayer ion-exchange yielding to the formation of a nanocomposite. 3.2.3. Elastomeric nanocomposites In a first study, Lan and Pinnavaia [71] have examined the formation of nanocomposites with a rubber-epoxy matrix obtained from a DGEBA derivative (Epon 828) cured with a polyether diamine (Jeffamine D2000) so as to reach subambient glass transition temperatures. Two montmorillonites modified, respectively, by the protonated n-octylamine and the protonated n-octadecylamine have been used in this study. It has been shown that depending on the alkyl chains length of modified clays, an intercalated and partially exfoliated (n-octyl) or a totally exfoliated (n-octadecyl) nanocomposite can be obtained. With the protonated octadecylamine-modified clay, heating the reaction mixture at 758C triggered the epoxide and diamine to migrate into the clay galleries and to Ê . Upon additional heating, further form an intermediate with an interlayer spacing of 54 A polymerization occurs (catalyzed by the protonated primary amine) with deep penetration of the components within the gallery, leading to the formation of the exfoliated structure. In the case of the octyl derivative whose hydrophobicity is lower, the amount of intercalated epoxide and diamine is insufficient to achieve complete exfoliation. Therefore, only a portion of the clay is delaminated, as Ê reflection, evidenced by the broadening and the decrease in the scattering intensity of the 15.2 A corresponding to the modified clay interlayer spacing. These authors have also studied other parameters such as the nature of alkylammonium cations present in the gallery and the effect of the cation-exchange capacity of the clay [72] when Epon 828 was cured with m-phenylene diamine. It was demonstrated that when mixed with the epoxide and whatever the length of the protonated primary alkylamine (with 8, 10, 12, 16 or 18 carbons), the modified clays adopt a structure where the carbon chains are fully extended and oriented perpendicularly to the silicate plane incorporating maximum one monolayer of epoxide molecules. In order to activate clay exfoliation, presence of acidic species in the intergallery is necessary, as illustrated (Fig. 20) by the decrease in layer exfoliation with decreasing BroÈnsted acidity of the exchange ion in the order CH3(CH2)17NH3‡ >CH3(CH2)17N(CH3)H2‡>CH3(CH2)17N(CH3)2H‡>CH3(CH2)17N(CH3)3‡. The effect of cationic exchange capacity has been addressed using four different clays (hectorite, CECˆ67 meq/100 g; montmorillonite, 90 meq/100 g; fluorohectorite, 122 meq/100 g; vermiculite, 200 meq/100 g) modified by CH3(CH2)15NH3‡. XRD patterns of the diamine-cured

25

26

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 20. XRD patterns of diamine-cured epoxy/clay nanocomposites formed from montmorillonite clays (5 wt.%) containing primary, secondary, tertiary and quaternary onium ions with n-C18 chain lengths (reproduced from [72] with permission).

epoxy clay composites formed with 5 wt.% loading of the modified clays show that an exfoliated structure is formed with hectorite and montmorillonite while with fluorohectorite, a partially exfoliated/partially intercalated structure is obtained. For vermiculite, an intercalated structure is obtained because the density of onium ions present is very high and limits the penetration of monomers. As a result, only a little amount of epoxy resin is formed inside the intergallery that cannot expand enough to undergo exfoliation. The same kind of results were obtained by Zilg et al. [73] who cured DGEBA with hexahydrophtalic acid anhydride in the presence of different clays (fluoromica, hectorite and bentonite) modified by a wide variety of oniums. An original result comes from the inefficiency of functionalized oniums (carboxylic acid or hydroxyl group) to improve exfoliation comparing to simple protonated alkylamine. Starting from another type of layered material, magadiite, which is a layered silicic acid, Wang and Pinnavaia [74] have developed a different type of exfoliated structure where the layers were still Ê of elastomeric epoxy matrix arising from the curing of Epon organized but spaced by around 80 A 828 (DGEBA derivative) by Jeffamine 2000 (diamine polyether). Such unusual structure can be obtained by carefully controlling of the organo-modification of magadiite with a mixture of alkylammonium cation and the corresponding alkylamine (dimethyloctadecylamine) in order to form a paraffin-like environment between the silicic acid layers. It has to be noted that exfoliation of magadiite layers within the epoxy matrix can only be achieved with this paraffin-like structure. If the alkylamine is not used, a `lateral monolayer' is formed which is unable to swell in the monomer slurry and does not lead to the formation of a nanocomposite. Wang and Pinnavaia [75] have also synthesized intercalated nanocomposites based on elastomeric polyurethanes. An organomontmorillonite modified with the protonated dodecylamine or octadecylamine is swollen in a polyol such as ethylene glycol, poly(ethylene glycol), or Voranol (glycerol propoxylate with molecular weight ranging from 700 to 3000), then cross-linked using a commercial methylene diphenyl diisocyanate prepolymer (Rubinate). After curing at 508C for 12 h Ê. an intercalated nanocomposite can be obtained with an interlayer spacing of 50.8 A

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 21. Variation of entropy change per area unit (hDSV/NAkB) in function of the change in gallery height for an arbitrary polymer and a silicate functionalized with octadecylammonium groups calculated for the polymer chain (dashed point line), the tethered chains (dashed line) and addition of the two contributions (plain line). h1ÿh0 is the change in gallery height for a fully extended octadecyl chain (reproduced from [76] with permission).

3.3. Melt intercalation 3.3.1. Melt intercalation of modified montmorillonite 3.3.1.1. Theoretical concepts. The thermodynamics that drives the intercalation of a polymer inside a modified layered silicate while the polymer is in the molten state has been approached through a lattice-based mean field theory by Vaia and Giannelis [76]. They found that, in general, the outcome of polymer intercalation is determined by an interplay of entropic and enthalpic factors. In fact, although the confinement of the polymer chains inside the silicate galleries results in a decrease in the overall entropy of the macromolecular chains, this entropic penalty may be compensated by the increase in conformational freedom of the tethered alkyl surfactant chains as the inorganic layers separate, due to the less confined environment as depicted in Fig. 21. Since small increases in the gallery spacing do not influence strongly the total entropy change, intercalation will rather be driven by the changes in total enthalpy. In this study, the enthalpy of mixing has been classified in two components: apolar interactions generally unfavorable and arising from interaction between polymer and surfactant aliphatic (apolar) chains, and polar interactions which originate from the Lewis acid/Lewis base character of the layered polar silicates interacting with the polymer chains. Indeed, since in most conventional organo-modified silicates, the tethered surfactant chains are apolar, dispersion forces dominate polymer±surfactant interactions. On the other hand, a favorable energy decrease is associated with the establishment of many favorable polymer surface polar interactions. The enthalpy of mixing can thus be rendered favorable by maximizing the magnitude and number of favorable polymer±surface interactions while minimizing the magnitude and number of unfavorable apolar interactions between the polymer and the aliphatic chains introduced along the modified layer surfaces. Fig. 22 shows the free energy per area (hDfV) evolution in function of the gallery height (hÿh0) at 423 K for various esp,sa. The notation esp,sa corresponds to espÿesa where esp…or a† is the pair-wise silicate surface (s)±polymer (p) chain or surfactant (a) interaction energy per area of the s±p (or a) contacts relative to the initial s±s and p±p (or a±a) interaction. For the simplicity of the model, the pair-wise interaction between the polymer

27

28

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 22. Variation of the free energy per unit area (hDfV) in function of the change in gallery height (hÿh0) calculated for an arbitrary polymer interacting with a silicate layer functionalized with octadecyl ammonium groups. Curves are calculated for various values of the difference in pair-wise interaction energies between the silicate layer and the polymer chain and the silicate layer and the tethered chains esp,sa. The pair-wise interaction energy between the polymer chain and the tethered chains (eap) is taken equals to 0. Free energy curves I, IIa, IIb and III correspond to esp,sa values of 0, ÿ4, ÿ8, ÿ12 mJ/m2, respectively (reproduced from [76] with permission).

and the surfactant eap is taken equal to 0. This situation describes in fact melt intercalation of a series of polyethylene-based polymers with esp,saˆ0 for pure HDPE and esp,sa<0 for polyethylene copolymers in which a small ration of ethylene units are replaced with moieties exhibiting more favorable interactions with the silicate. The free energy curves may be grouped into three types. First, curves that are positive at all gallery heights (type I, esp,saˆ0). In this case, polymer intercalation is unfavorable, and the polymer and the organo-modified layered silicates are immiscible. The second type regroups the curves displaying one minimum (type IIa, esp,saˆÿ4) or more than one minimum (type IIb, esp,saˆÿ8) and corresponding, respectively, to well defined intercalated structure and ill-defined intercalated structures or intermediate intercalated structures before complete layer exfoliation. Finally, the third type of curves displays a continuous decrease in the free energy values with gallery height expansion (type III, esp,saˆÿ12) indicating that polymer intercalation and complete layer separation is favorable. This last type corresponds to complete polymer-silicate layer miscibility, characteristic of exfoliation. Balazs et al. [77,78] have considered the self-consistent field theory (SFC) in order to investigate the factors promoting the penetration of polymers into layered silicates. They first varied properties related to the nature of the tethered surfactant chains. They found out that an increase in the surfactant length (approaching the length of the polymer chains) improves the layers separation by the formation of a broad interface (or interphase) which allows the polymer from adopting more conformational degrees of freedom. Accordingly, exfoliated or intercalated structures can be formed even for slightly unfavorable interactions between the polymer and the modified surfaces. On the opposite side, increase in the length of polymer chains tends to render the interlayer mixture immiscible. These authors also reported on the effect of the surfactant density on the intercalation process, showing that excessive density of tethered alkyl chains can impede the formation of intercalated structures. In order to model the macroscopic phase behavior of the polymer/clay mixture, Balazs et al. [78,79] have adapted to disk-like particles the Onsager model for the equilibrium behavior of rigid

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 23. Phase diagrams of disks dispersed in a polymer matrix for different disk aspect ratios. The disk aspect ratios (D/L) were varied by changing the dimension of the diameter of the disks (D), keeping their thickness constant (Lˆ1) (reproduced from [79] with permission).

rods set up dispersed in a polymer matrix. From this new model, one can calculate phase diagrams of the polymer/clay composites in function of the Flory±Huggins interaction parameter. These diagrams can differentiate immiscible and miscible regions further separated in isotropic or nematic (relatively ordered) arrangement. In this study, the blend miscibility appears to be strongly negatively influenced by an increase in the length of polymer chains. For very high polymer length, the particle/ polymer mixture gets immiscible even for negative values of the interaction parameter. The phase diagram appears to be also strongly dependent upon the aspect ratio of the particle (i.e. the diameter (D) to thickness (L) ratio of the assumed disk-shaped particle) as shown in Fig. 23. An increase in the particle diameter favors the apparition of a nematic phase at a low volume fraction implying the formation of relatively ordered structures for low clay particle content. Finally, Ginzburg and Balazs [80] have developed an even more complex model, based on perturbation-type density functional theory to describe the complete (isotropic, nematic, smectic A, columnar and intercalated called `crystalline') phase diagram of an incompressible polymer disk mixture. The phase diagram is shown to be strongly dependent upon the shape anisotropy of the filler particles, the polymer chain length, and the strength of the interparticle interaction. For instance, an increase in the interparticle interaction strength leads to the complete disappearance of the nematic phase in favor to direct coexistence between isotropic and columnar or `crystalline' (intercalated) structure. 3.3.1.2. A model polymer: polystyrene. Melt intercalation of polystyrene and derivatives has been widely studied and has served to experimentally describe the aforementioned thermodynamics as well as the kinetics and morphologies accordingly generated. Vaia and Giannelis [81] have studied PS as the matrix for dispersing different types of clays. Lifluorohectorite (CECˆ150 meq/100 g), saponite (100 meq/100 g), and sodium montmorillonite (80 meq/100 g) were accordingly modified using various ammonium cations: dioctadecyldimethylammonium, octadecyltrimethylammonium, and a series of primary alkylammonium cations with carbon chains of 6, 9±16 and 18 carbon atoms. The nanocomposites were synthesized by statically annealing (without any mixing or shearing) a pelletized intimate mix of the modified silicate in PS (Mwˆ30,000, Mw/Mnˆ1.06) under vacuum at 1708C, a temperature well above the PS glass transition [82] (Table 10). Comparison of the first three entries indicates that, for a given alkyl surfactant, increasing the cation exchange capacity from 80 to 120 meq/100 g allows for PS intercalation to occur. At a low CEC and for single alkyl chain built up cation (entries 1 and 2), no intercalation is observed. Under these conditions, aliphatic alkyl chain of the organic cation adopts a pseudo monolayer arrangement

29

30

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 10 Characteristics of polystyrene melt intercalation within octadecylammonium modified claysa Entry

Clay (CEC in meq/100 g)

Ammonium cation

Initial gallery height (nm)

Final gallery height (nm)

Net change (nm)

1 2 3 4 5 6 7

M (80) S (100) F (120) F (120) M (80) S (100) F (120)

PODAb PODAb PODAb QODAc DODMDAd DODMDAd DODMDAd

0.75 0.83 1.33 1.57 1.43 1.50 2.85

0.75 0.83 2.16 2.69 2.25 2.35 2.85

0 0 0.83 1.12 0.82 0.85 0

a

Mˆmontmorillonite, Sˆsaponite and Fˆfluorohectorite. PODA: primary octadecylammonium. c QODA: quaternized octadecylammonium. d DMDODA: dioctadecyldimethylammonium. b

characterized by low gallery height. Interestingly enough, intercalation can take place at lower CECs at the condition to modify the clay surface with ammonium cation bearing two long alkyl chains (entries 5 and 6). However, excessive packing of the chains all along the layer surface (high CEC and two long alkyl chains per cationic head, entry 7) leads to the formation of a non-intercalated structure as predicted by the theory introduced by Vaia and Giannelis [76] (vide supra). The authors also found out that polymer intercalation depends on the length of the exchange ammonium cation as well as on the annealing temperature. At annealing temperature equal to or lower than 1608C and for chain lengths inferior to 12 carbon atoms, no intercalation is detected, while for higher chain lengths PS readily intercalates. At 1808C, intercalation occurs whatever the ammonium chain length be. However, the full width at half-maximum of the XRD diffraction peak, that is known to witness for the intercalation regularity, increases for chain lengths lower than or equal to 12 carbon atoms. Such a peak broadening attests for a more disordered and irregular structure and arises only for organomodified clays with a pseudo bilayer chain arrangement. The authors interpret this behavior as an extended propensity of these organoclays to separate under these conditions, exfoliation being nevertheless kinetically limited by the static conditions used to produce the nanocomposites, i.e. simple annealing without mixing or shearing. The effect of PS molecular weight was also studied. Contrary to what was predicted by the theory, intercalation occurred whatever the Mw of the PS chains. Only the period of time needed for the intercalation to proceed was different, going from 6 h for a Mw of 30,000 to 24 h for 90,000 and 48 h for 400,000 at 1608C. Clearly, high molecular weight PS chains decrease the kinetics of intercalation by decreasing the diffusivity of the polymer in the interlayer. Finally, the authors studied the influence of the nature of the polymer matrix, creating under the same experimental conditions different composites based on poly(vinylcyclohexane) (PVCH), poly(3-bromostyrene) (PS3Br) and poly(vinylpyridine) (PVP). All these polymers appear to be immiscible with the dioctadecyldimethylammonium-modified fluorohectorite but interestingly formed intercalated structures with montmorillonite (thus with a lower CEC) modified by the same ammonium cation with an exception for PVCH. These differences in intercalation behavior indicate that the chemical nature of the polymer pendant group (polar or not) greatly affects the formation of the hybrid composition. Moreover, in the case of intercalated PVP, a broadening of the diffraction peak in XRD pattern indicated a less regular arrangement of the intercalated structure possibly due to some propensity for exfoliation. As far as the dodecylammonium-modified fluorohectorite was concerned, the broadening of the diffraction peak observed for PVP-based nanocomposite was even more pronounced attesting for a very disordered structure. These results are totally in accordance

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 24. Temporal series of XRD patterns for a organo-modified fluorohectorite/PS pellet annealed in situ at 1608C in vacuum. p(0 0 1) and p(0 0 2) locate the basal reflections for the unintercalated fluorohectorite while i(0 0 1), i(0 0 2) and i(0 0 3) correspond to the basal reflections for the intercalated nanocomposite that is forming with time (reproduced from [83] with permission).

with the proposed thermodynamic model that anticipated a better intercalation and even exfoliation when polar interactions between the layered silicates and the polymer chains are enhanced. Vaia et al. [83] have also studied the kinetics of melt intercalation by following the time evolution of XRD diffraction patterns for statically annealed polystyrene/octadecylammoniumexchanged fluorohectorite. The change in intensity of the pristine and intercalated diffraction peaks with time was used as a reflect of the kinetics of the polymer intercalation process (Fig. 24). By integrating the intensity of both non-intercalated and intercalated peaks, the authors were able to estimate the fraction of intercalated silicates as a function of the annealing time. Influence of both annealing temperature and PS molecular weight were then determined. Higher annealing temperatures, as well as lower molecular weights, increase the rate of PS intercalation. In order to give a more concrete interpretation of the kinetic data, the morphology of the modified montmorillonite has been examined by TEM. The authors found that the nominal silicate particle (agglomerate of ca. 175 mm in diameter) consist of smaller oblong-shaped particles, coined as primary particles forming agglomerates. Those primary particles of 1±10 mm length themselves consist of a compact face-to-face stacking or low-angle intergrowth of individual silicate crystallites (also known as tactoids). These crystallites are built up as a coherent stacking of individual silicate layers.The layers are roughly circular, 0.05±0.5 mm in diameter, and ca. 1 nm thick. They are separated by a van der Waals interlayer (or gallery), which contains the alkylammonium cations Ê thick) in the pristine organosilicate. This overall morphology can be schematized as shown (10 A in Fig. 25. Based on this morphology and TEM observations of partially intercalated composites, it has been demonstrated that the accessibility of the interlayer to the polymer chains depends on the location and orientation of the primary particles within the agglomerates and on the location and orientation of the crystallites within the primary particles, meaning that crystallites near the edge will be more accessible to polymer chains than those near the center. Since the silicate layers are impenetrable, the polymer must enter the gallery from the edges of the crystallites. The authors observed that, under their experimental conditions, the polymer penetrates the agglomerate and surrounds the primary particles before the occurrence of substantial intercalation. Therefore, they assumed that the polymer penetration within the agglomerate was not a limiting step for

31

32

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 25. Schematic of the morphology of organo-modified fluorohectorite (reproduced from [83] with permission).

intercalation. Furthermore, they demonstrated that the rate of PS intercalation was dependent upon the size of the primary particles, bigger primary particles being less intercalated than smaller ones, for a given annealing time. This size dependence was explained by the lack of significative difference in the mass transport in between and through the crystallites within a given primary particle so as polymer intercalation could be described as a Fickian process with a single, apparent diffusivity. The polymer diffusion into the primary particles can be sketched as a mass flow through the curved lateral surface of a cylinder bearing impermeable flat circular faces of area equal to the mean primary particle size. Accordingly, the activation energy of the melt intercalation was calculated based on the experimental values of the effective diffusional rate obtained at different temperature. Assuming an Arrhenius temperature dependence, the activation energy for the PS (Mwˆ30,000 and Mw/Mnˆ1.06)/modified fluorohectorite composite is 16612 kJ/mol, comparable to the activation energy measured for self-diffusion of PS (167 kJ/mol) [83]. This indicates that the largest energy barrier to PS intercalation is comparable to that of PS chain motion within the polymer melt. Since mass transport into the primary particle has been observed to be the limiting step to complete intercalation, any dynamic mixing should decrease the intercalation time by breaking down the primary particles and increasing the sample uniformity, that opens the way to the preparation of intercalated PS within a few minutes. In a further study, Vaia et al. [84] have synthesized, by the above reported static annealing procedure, ordered and disordered intercalated nanocomposites by mixing, respectively, 30,000 or 400,000 Mw PS with octadecylammonium-exchanged fluorohectorite on one side and 30,000 Mw PS or 55,000 Mw poly(3-bromostyrene) with dodecylammonium-exchanged fluorohectorite on the other side. TEM characterization shows that the ordered intercalates exhibit regular microstructures similar to the non-intercalated organo-modified fluorohectorite. In this case, polymer intercalation occurs as a front which penetrates the primary particles from the external edges while for disordered intercalated systems, a heterogeneous microstructure is observed, with pronounced interlayer disorder with greater spacings towards the polymer-primary particle boundary. In these latter nanocomposites, individual silicate layers are observed near the edges, whereas small coherent layer packets separated by polymer-filled gaps are prevalent inside the primary particle. This study also

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

shows that intercalation could be favored by the presence of defects in the layer stacking that would allow the polymer to penetrate more easily inside the interlayer spaces. 3.3.1.3. Nylon-6. While the preparation of nanocomposites based on nylon-6 matrix has been widely described using the in situ intercalative polymerization (see Section 3.2.1), less attention has been drawn to the nylon-6-based nanocomposites prepared by melt blending. Liu et al. [85] have prepared nanocomposites based on a commercial nylon-6 melt blended with an octadecylammonium-exchanged montmorillonite (CECˆ100 meq/100 g) in a twin screw extruder. They prepared composites with a filler content ranging from 1 to 18 wt.%. An intercalated structure was observed to be formed by XRD for composites containing more than 10 wt.% of the organoclay, with Ê for the pristine organoclay to 36.8 A Ê for the intercalated an interlayer spacing increasing from 15.5 A species. At filler content lower than 10 wt.%, no interlayer spacing could be detected through XRD and the TEM micrographs allow for the observation of an exfoliated structure, indicating that exfoliation in this case is highly dependent upon the filler content. XRD and DSC data also showed that exfoliated structures strongly influenced the nature of the nylon-6 crystallization, favoring the formation of g-crystals in addition to the crystals of the a-form observed in the native nylon-6 matrix. Moreover, DSC cooling scans showed that exfoliated layered silicates highly increased the crystallization rate, having a strong heterophase nucleation effect. 3.3.1.4. Polypropylene. Polypropylene (PP) has also been tested for the preparation of nanocomposites. However, no direct intercalation of PP in simply organically modified layered silicates has been observed so far, PP being too much apolar to correctly interact with the modified layers. In a first study, Kato et al. [86] described the melt intercalation of PP chains modified with either maleic anhydride (PP-MA) or hydroxyl groups (PP-OH) in octadecylammonium-exchanged montmorillonite (CEC: 119 meq/100 g). When PP-MA (Mwˆ30,000, acid valueˆ52 mg KOH/g) or PP-OH (Mwˆ20,000, OH valueˆ54 mg KOH/g) was melt blended under shearing with a same given amount of modified montmorillonite at 2008C for 15 min, intercalated nanocomposites were Ê for the initially organorecovered as demonstrated by the increase in the layer spacing from 21.7 A Ê for PP-MA and PP-OH based nanocomposites, modified montmorillonite to 38.2 and 44.0 A respectively. Interestingly enough, a PP-MA matrix with a lower maleic anhydride content (acid valueˆ7 mg KOH/g for Mwˆ12,000) did not intercalate under the same conditions, showing that a minimal functionalization of the PP chains has to be reached for intercalation to proceed. The authors also examined the effect of polymer to clay ratio on the intercalation extent and showed that Ê for a PP-MA to intercalation increased when the PP-MA fraction was increased, going up to 72.2 A clay ratio of 3:1. Intercalation of PP-MA in modified clay was used in order to prepare PP-based nanocomposites [87,88]. In both studies, the three components (PP, PP-MA and modified clay) were melt blended in a twin-screw extruder at 2108C in order to obtain composites filled with 5 wt.% clay. Formation of an exfoliated structure was observed for: (1) relatively high PP-MA content (typically 22 wt.%), (2) sufficient polar functionalization of PP-MA chains (acid valueˆ26 mg KOH/g for Mwˆ40,000). However, the relative content in maleic anhydride cannot exceed a given value in order to keep some miscibility between PP-MA and PP chains. Indeed, when too many carboxyl groups are spread along the polyolefin chains (e.g. acid valueˆ52 mg KOH/g), no further increase in the interlayer spacing was obtained in clay/PP/PP-MA blends, leading rather to the dispersion of PP-MA intercalated clay in the PP matrix. Another way to obtain nanocomposites from organo-modified clays and PP has been recently proposed by Wolf et al. [89]. In this technique, the authors modified a commercially available

33

34

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 11 Interlayer spacing of various modified montmorillonite and the resulting composites obtained with EVA (10.76 mol% vinyl acetate) Code Mont-Na Mont-2CN2C18 Mont-NC11COOH Mont-3CNC21COOH

Cation ‡

Na (CH3)2N‡(C18H37)2 H3N‡C11H22COOH (CH3)3N‡C21H42COOH

Ê) Interlayer spacing (A In modified clay

In EVA composite

12.2 31.9 16.3 20.1

12.6 40.3 16.7 20.1

organoammonium-exchanged montmorillonite using an organic swelling agent (whose boiling point is situated between 100 and 2008C, such as ethylene glycol, naphtha or heptane) in order to increase the interlayer spacing. The swollen organo-modified clay was then compounded with PP in a twinscrew extruder at 2508C. The swelling agent was volatized during extrusion process, leading to the Ê formation of a `nano' composite which did not present any crystalline reflection in the 20±40 A range of XRD patterns. 3.3.1.5. Ethylene-vinyl acetate copolymers. In a very recent work, ethylene-vinyl acetate copolymers with various vinyl acetate contents (4.2, 7.1, 10.8 and 24.8 mol% vinyl acetate) have been used as matrices for the preparation of nanocomposites [90]. The presence of polar groups (ester group of the vinyl acetate moieties) all along the chains improves the ability of these copolymers to intercalate in organo-modified montmorillonites. Several exchanging cations bearing either simple alkyl chains or aliphatic chains terminated by a carboxylic group have been studied for modifying montmorillonites and are described in Table 11. Nanocomposites were only formed when EVA copolymers were melt blended at 1308C with non-functionalized organo-montmorillonites such as montmorillonite exchanged with dimethyldioctadecyl ammonium (Mont-2CN2C18). A partially intercalated±partially exfoliated structure was observed by both the presence of peaks characteristic of the intercalation process in the XRD patterns (see Table 11) and appearance of dispersed silicate layers in TEM micrographs (Fig. 26). This intercalation±exfoliation morphology occurs even at low vinyl acetate content (4.2 mol%) in the copolymer matrix while no intercalation is observed for HDPE, thus in the absence of polar ester groups. It is moreover independent of the processing temperature. A set of experiments based on the EVA matrix containing 10.8 wt.% of vinyl acetate has also allowed to determine that concomitant intercalation and exfoliation occur whatever the filler content (from 1 to 50 wt.% of Mont-2CN2C18) even if the exfoliation step tends to decrease when the organo-montmorillonite amount increases. With the same EVA matrix, the use of ammonium cations functionalized with carboxylic groups did not lead to the formation of an intercalated structure (see last two entries in Table 11), indicating that the functionalization of the clay interlayer is detrimental to the intercalation process. EVA copolymers appears to easily form nanocomposites even if totally exfoliated structures have not been achieved yet. 3.3.1.6. Poly(styrene-b-butadiene) copolymer (SBS). Symmetric (styrene±butadiene±styrene) block copolymer has been the first thermoplastic elastomer matrix investigated in the preparation of nanocomposites. Laus et al. [91] have melt blended a commercial SBS (Mnˆ70,000, Mw/Mnˆ1.18, and 30 wt.% in PS) with a dimethyldioctadecylammonium-exchanged montmorillonite in a Brabender at 1208C for 10 min. Composites with 10, 20 and 30 wt.% of filler were compounded. A comparative set of composites with the corresponding native Na-montmorillonite has also been

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 26. TEM micrograph of organo-modified montmorillonite exfoliated in a ethylene-vinyl acetate copolymer matrix.

prepared. In addition, all the samples were further annealed at 1208C under nitrogen for time periods ranging from 16 to 73 h in order to study the effect of the thermal treatment on their structural characteristics. The authors have controlled that no chemical modification or degradation occurred during compounding and thermal treatments by checking the molecular parameters of the SBS copolymers. SBS recovered by complete extraction in CHCl3 of the so prepared composites displayed unmodified molecular parameters compared to starting materials. While no change in XRD patterns appears for the composites prepared with Na-montmorillonite (interlayer spacingˆ Ê for the filler and both three composites), a substantial change in the XRD patterns could be 12.7 A observed for the composites based on the organo-montmorillonite. These composites showed both a Ê for the organoclays to 46 A Ê in the composites) small increase in the interlayer spacing (from 41 A and a broadening of the diffraction peak that increased with annealing time. These information tend to indicate the formation of intercalated nanocomposites that is far from being complete during the blending time. Analysis of the dynamic mechanical properties (see also Section 4.1.3) shows an increase in the glass transition temperature (Tg) of the poly(styrene) outer blocks (PS) with the filler content while the Tg corresponding to the poly(butadiene) block (PBD) is kept unchanged (Fig. 27). Under very similar conditions, the presence of pristine Na-montmorillonite does not have any effect on both PS and PBD blocks. This Tg increase for the PS blocks (which is also enhanced for longer annealing times) can be interpreted by a selective intercalation of the styrene blocks into the silicate galleries, which is more important at higher nanofiller content and longer annealing times. 3.3.1.7. Elastomers. Burnside and Giannelis [92] have described the two-step preparation of silicon rubber-based nanocomposites. First, silanol-terminated poly(dimethylsiloxane) (PDMS, Mwˆ 18,000) was melt blended at room temperature with dimethylditallowammonium-exchanged montmorillonite then the silanol end groups were cross-linked with tetraethylorthosilicate (TEOS) in the presence of tin bis(2-ethylhexanoate) as catalyst at room temperature. However, in order to

35

36

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 27. Trends of the glass transition temperatures of the polystyrene (squares) and polybutadiene (circles) blocks domains for symmetric (styrene±butadiene±styrene) block copolymer-based nanocomposites (open symbols) and microcomposites (full symbols) as a function of the filler content (reproduced from [91] with permission).

obtain exfoliated nanocomposites (as determined by featureless XRD patterns), several conditions were required such as mixing the modified clay and PDMS under sonication and addition of a small quantity of water (typically corresponding to a monolayer coverage of the silicate surface). Nature of both silicon matrix and clay modifier play an important role. For example, neither exfoliation nor intercalation can occur if montmorillonite is modified with benzyl dimethyloctadecylammonium cation or if a too large water amount is added. On another side, only intercalation was observed when a PDMS-poly(diphenylsiloxane) random copolymer containing 14±18 mol% diphenylsiloxane units was used. The results stress out again the key importance to get a right match between matrix and organoclay in order to optimize the layered silicate exfoliation. More recently, Wang et al. [93] have prepared a series of intercalated nanocomposites based on the same type of silicon rubber. In their synthesis, they dispersed a hexadecyltrimethylammoniumexchanged montmorillonite in a silanol-terminated PDMS (Mwˆ68,000). After heating the dispersion at 908C for 8 h, TEOS and dibutyltin dilaurate were added and the materials were further cured at room temperature for 12 h. Under these conditions, a silicone rubber-based nanocomposite was obtained with an intercalated structure as determined by the increase of the Ê (organoclay) to 37.1 A Ê for the composites. TEM observation interlayer spacing from 20.2 A confirms the intercalation but also indicates that the filler is uniformly dispersed in the rubber matrix as crystallites about 50 nm thick. Okada and co-workers [94,95] obtained a nitrile rubber (NBR)-based nanocomposite in a twostep synthesis. They first modified a Na-montmorillonite through cation exchange with an aminoend-capped poly(butadiene-acrylonitrile) oligomer (Mwˆ3400) cationized by HCl in water. This modified clay was then melt blended on a two-roll mill with NBR and usual additives for vulcanization such as sulfur and ZnO were added in order to obtain vulcanized rubber sheets after compression molding at 1608C for 15 min. Even if no direct and objective evidence of the `nano' structure is given, a large number of properties (gas permeability, enhanced mechanical properties, . . .) tends to demonstrate that the behavior of these NBR-based composites is in the range of what is usually observed for nanocomposites. 3.3.2. Melt intercalation of unmodified montmorillonite Poly(ethylene oxide) (PEO), is a polymer that readily intercalates by melt blending in pristine ‡ Li or Na‡ montmorillonite. Vaia et al. [96] have prepared PEO-based intercalated nanocomposites

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 28. DSC traces for poly(ethylene oxide)/Na-montmorillonite mixtures heated to 808C for 0, 2 and 6 h (reproduced from [96] with permission).

by `statically' annealing at 808C a pelletized mixture of PEO and Li‡ (or Na‡) montmorillonite (CECˆ80 meq/100 g). Under these conditions, complete intercalation arises within 6 h, increasing Ê (for the pristine montmorillonite) to 17.7 A Ê for the intercalated the interlayer spacing from 11.4 A species. It has to be noted that this last value corresponds exactly to what has been obtained through preparation by the exfoliation±adsorption method. The intercalation is further observed through FTIR measurements where the C±H stretching vibration within the PEO chain at 2900 cmÿ1 is split into a doublet at 2910 and 2878 cmÿ1 due to polymer host interactions. Finally, the disappearance of the melting transition of PEO after a 6 h annealing at 808C (Fig. 28) indicates that the intercalation of PEO leads to a confinement of the macromolecules that prevents the polymer from crystallizing. In addition to the melting enthalpy decrease, the melting endotherm also shifts towards lower temperatures. The authors interpret this phenomenon by the interaction of water molecules (displaced from the interlayer by the PEO intercalation) with `external' PEO crystallites. Besides studies dealing with the relaxation of confined chains through either the evolution of the intercalated PEO glass transition properties [97] or conformation by NMR techniques [10], the PEO intercalation in unmodified montmorillonite has been interestingly exploited to favor the co-intercalation of poly(methyl methacrylate) [98]. This co-intercalation is claimed to be a potential way to enhance the ionic conductivity of polymer-layered silicate nanocomposites (see also Section 4.4.1). A theoretical approach to the intercalation of macromolecules in unmodified montmorillonite has been described by Balazs et al. [77] using both numerical and analytical self-consistent field theory. They showed that it is theoretically possible to promote the exfoliation of an unmodified montmorillonite by melt blending it with a polymer mixture comprising the desired polymer matrix and a small amount of end-functionalized polymer whose terminal function could strongly interact with the silicate layers. This end-functionalized polymer could also be replaced by a diblock copolymer, one sequence of which can intercalate (like PEO) while the other one is either of the same nature than the polymer matrix or at least compatible with it. Under these conditions (but without taking into account kinetics factors), exfoliation would occur at a fraction of endfunctionalized polymer (or diblock copolymers) as low as 5 mol%. These theoretical results have found some experimental confirmations through the work of Fisher et al. [99] which describes the intercalation and exfoliation of unmodified layered silicates

37

38

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

(but also double layered hydroxides) by diblock copolymers comprising one block that can intercalates, e.g. PEO or poly(2-vinylpyridine), and a more hydrophobic block such as PS or PMMA. They first prepared an intercalated nanocomposite by `statically' annealing the layered silicate (hectorite, saponite or montmorillonite) with the diblock copolymer then, in a second step, partially exfoliated this nanocomposite by melt blending with a bulk polymer, the nature of which corresponds to the hydrophobic block of the copolymer. In the first step, intercalation is observed by XRD analysis where an increase in the interlayer spacing analogous to what is observed for pure PEO can be recorded. Moreover, depending on the relative length of the hydrophilic and hydrophobic sequences, a partially exfoliated nanocomposite can be formed as claimed by the authors for a diblock with a short PEO block (Mwˆ1000) and a longer PS block (Mwˆ3000). In the second step, the dispersion by extrusion of these intercalated or partially exfoliated nanocomposites in the corresponding hydrophobic polymer at least maintains the state of dispersion observed in the first step. This process, using unmodified layered silicates and commercially available diblock copolymers could be an alternative towards the need of carefully organo-modified clays. 3.4. Template synthesis A last technique reported for preparing layered silicate-based nanocomposites implies the in situ hydrothermal crystallization of the clay layers (hectorite) in an aqueous polymer gel medium where the polymers often act as template for the layers formation. This method is particularly adapted to water soluble polymers, and some attempts have been achieved with polymers such as poly(vinylpyrrolidone) (PVPyr), hydroxypropylmethylcellulose (HPMC), poly(acrylonitrile) (PAN), poly(dimethyldiallylammonium) (PDDA) and poly(aniline) (PANI) [100]. The typical method for in situ hydrothermal crystallization of a polymer/hectorite nanocomposite consists in refluxing for 2 days a 2 wt.% gel of silica sol, magnesium hydroxide sol, lithium fluoride and the desired polymer in water. XRD patterns attest for the formation of a polymer/hectorite intercalated nanocomposite. It is worth pointing out that the interlayer spacing linearly depends upon the wt.% of polymer incorporated as shown in Fig. 29 for PVPyr matrix. Delaminated structures are suspected in case of PANI and PAN, unusually weak peaks corresponding to the interlayer spacing are indeed observed in the XRD patterns. As far as PDDA is concerned, the polymer loading cannot exceed 20 wt.%. It seems that polymer incorporation within the growing layers are limited to the strict balance between negative charges of the clay layers and cations borne by the polymer chains. It has to be noted that the size of the layers obtained by this

Fig. 29. Correlation between d spacing from XRD patterns and weight percent polymer in synthetic PVP-hectorite clay nancomposites (reproduced from [100] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

template synthesis cannot compete with natural layered silicates for kinetic reasons and their average length is limited, at best, to about one-third of their natural counterparts. 4. Properties Layered silicate nanofillers have proved to trigger a tremendous properties improvement of the polymers in which they are dispersed. Amongst those properties, unexpected large increase in moduli (tensile or Young's modulus and flexural modulus) of nanocomposites at filler contents sometimes as low as 1 wt.% has drawn a lot of attention. Thermal stability and fire retardancy through char formation are other interesting and widely searched properties displayed by nanocomposites. Those new materials have also been studied and applied for their superior barrier properties against gas and vapor transmission. Finally, depending on the type of polymeric materials, they can also display interesting properties in the frame of ionic conductivity or thermal expansion control. 4.1. Mechanical properties 4.1.1. Effect on tensile properties 4.1.1.1. Young's modulus. The Young's modulus (or tensile modulus), expressing the stiffness of a material at the start of a tensile test, has shown to be strongly improved when nanocomposites are formed. Nylon-6 nanocomposites obtained through the intercalative ring opening polymerization of e-caprolactam (see Section 3.2.1), leading to the formation of exfoliated nanocomposites, show a drastic increase in the Young's modulus at rather low filler content (Table 12). Actually, the material stiffness is substantially enhanced whatever the way of preparation: polymerization within organo-modified montmorillonite (NCH) [18], polymerization within protonated e-caprolactam swollen montmorillonite (L-NCH) [50], and polymerization within natural montmorillonite, in the presence of e-caprolactam and an acid catalyst (one-pot-NCH) [51]. The dependence of Young's modulus measured at 1208C for exfoliated nylon-6 nanocomposites with various clay contents obtained by in situ intercalative polymerization of e-caprolactam in the Table 12 Effect of nylon-6-based nanocomposite preparation on the Young's modulus related to the filler content and the average molecular weight of the matrix Sample preparation

Filler content (wt.%)

MW (103)

Young's modulus (GPa)

Commercial nylon-6 NCCa NCHb c L-NCH d One-pot-NCH

0 5 4.7 5.3 4.1

13.0 13.0 16.3 19.7 22.6

1.11 1.06 1.87 2.04 2.25

a

NCC: montmorillonite-based nylon microcomposite. NCH: nanocomposite obtained by in situ intercalative polymerization of e-caprolactam in protonated aminododecanoic modified montmorillonite [18]. c L-NCH: nanocomposite obtained by in situ intercalative polymerization of e-caprolactam in protonated e-caprolactam modified montmorillonite [50]. d One-pot-NCH: nanocomposite obtained by in situ intercalative polymerization of e-caprolactam with Namontmorillonite [51]. b

39

40

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 30. Dependence of tensile modulus E at 1208C on clay content for organo-modified montmorillonite and saponitebased nanocomposites (reproduced from [18] with permission).

Ê) presence of protonated aminododecanoic acid-modified montmorillonite (average length: 1000 A Ê and saponite (500 A) [18] is shown in Fig. 30. This dependence clearly indicates that the ability of dispersed silicate layers to increase the Young's modulus of nylon-6-based nanocomposites can be directly related to the average length of the layers, hence to the aspect ratio of the dispersed nanoparticles. Moreover, the difference in the extent of exfoliation, as observed for nylon-6-based nanocomposites synthesized by in situ intercalative polymerization of e-caprolactam using Na-montmorillonite and various acids [51], strongly influences the measured Young's modulus values (Table 13). Depending on the nature of the acid added to catalyze the polymerization, one can observe variation of the XRD peak intensity (Im) that is inversely related to the amount of exfoliated layers in the nanocomposite. For an increase in the Im values, a parallel decrease in the Young's modulus is observed, indicating that exfoliated layers are the main factor responsible for the stiffness improvement, while intercalated particles, having a less important aspect ratio, rather play a minor role. All these observations are furthermore confirmed in Fig. 31 which presents the evolution of the Young's modulus at room temperature of nylon-6 nanocomposites obtained by melt intercalation in function of the filler weight content measured in this case at room temperature [85]. The preparation of nanocomposites by this technique has the advantage to use the same matrix for each composite, thus with the same Mw and MWD nylon-6. Fig. 31 shows a constant and large increase in the modulus up to ca. 10 wt.% of nanoclay, above this threshold the Young's modulus seems to level off. This change exactly corresponds to the passage from totally exfoliated structure (below 10 wt.%) to partially exfoliated±partially intercalated structure (for 10 wt.% and upper) as determined by XRD and TEM analyses [85]. The same behavior can account for the evolution of Young's modulus in polypropylene nanocomposites obtained by melt intercalation when the amount of maleic anhydride-modified PP Table 13 XRD peak intensity (Im) and Young's modulus of various nylon-6-based nanocomposites obtained by a one-step in situ intercalative polymerization of e-caprolactam with Na-montmorillonite in the presence of different acids Acid

Im (cps)

Young's modulus (GPa)

Phosphoric acid Hydrochloric acid Isophtalic acid Benzenesulfonic acid Acetic acid Trichloroacetic acid

0 200 255 280 555 585

2.25 2.05 1.74 1.74 1.63 1.67

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 31. Effect of clay content on tensile modulus, measured at room temperature, of organomodified montmorillonite/ nylon-6-based nanocomposite obtained by melt intercalation (reproduced from [85] with permission).

(PP-MA) added to increase intercalation (and to possibly favor exfoliation, see Section 3.3.1) is varied [88]. Results are reported in Table 14 and compared to the corresponding microcomposite as well as to simple PP-MA/PP polymer blends. One can immediately see that increasing the amount of PP-MA (from sample PPCH 1/1 to PPCH 1/3) not only improves intercalation or partial exfoliation, but increases also the modulus value. Comparison of PP with the simple PP-MA/PP blends rules out any possible effect of some matrix modification due to the presence of increasing amounts of PP-MA (entries 1±3, Table 14). EVA-based nanocomposites obtained by melt intercalation within dimethyldioctadecylammonium modified montmorillonite [90] also display a non-linear evolution of the relative Young's modulus with the filler content (vol.%) (Fig. 32). As XRD and TEM observations for each filler content indicate that both intercalation and exfoliation occur, the non-linear increase in the relative tensile modulus may be explained by a decrease in exfoliated particle fraction at higher filler content. Another possible explanation would take into account a continuous variation of the mean aspect ratio of the primary particles, decreasing when the filler content is increased. This explanation is supported by the way the experimental points in Fig. 32 follow the theoretical curves calculated for various aspect ratios, i.e. f, from 12.5 to 20, using the modified Guth model [101], initially studied to describe the evolution of Young's modulus in true elastomeric matrices filled with carbon black particles that organize in high aspect ratio structures. Actually, at very low filler content (volume fractionˆ0.05, that corresponds, to 1 wt.%), experimental values for the relative tensile modulus range above the higher theoretical curve (drawn Table 14 Influence of maleic anhydride-modified polypropylene content on the stiffness of PP matrices and PP/clay nanocompositesa Sample

Filler content (wt.%)

PP-MA content (wt.%)

Young's modulus (MPa)

PP PP/PP-MA 7 PP/PP-MA 22 PPCC PPCH 1/1 PPCH 1/2 PPCH 1/3

0 0 0 6.9 7.2 7.2 7.2

0 7.2 21.6 0 7.2 14.4 21.6

780 714 760 830 838 964 1010

a

PPˆpolypropylene; PP-MA x: polypropylene modified by maleic anhydride (xˆwt.% of PP-MA in the blend); PPCCˆpolypropylene-based microcomposite; PPCH y/zˆpolypropylene-based nanocomposite (y/zˆweight ratio between y parts of filler and z parts of PP-MA).

41

42

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 32. Evolution of the relative Young's modulus (Ey (nanocomposite)/Ey (unfilled EVA)) with the filler volume fraction and confrontation of the experimental results to the modified Guth model for different aspect ratios. Modified Guth model: Ey (composite)/Ey (matrix)ˆ1‡0.67Ffrv‡1.62Ff2 r2v where Ey is the Young's modulus, rv the filler volume fraction and Ff is the aspect ratio of the filler particles.

for fˆ20). Then, at higher filler content, they level off and fit curves calculated for much lower aspect ratios. This suggests that silicate nanolayers with very high shape factor are only predominant at very low filler content. In contrast to the above results, when simply intercalated structures (without any exfoliation) are concerned, such as for PMMA [43] or PS [45] based nanocomposites, obtained by emulsion polymerization in presence of water-swollen Na-montmorillonite, the increase in Young's modulus is relatively weak, going, e.g. from 1.21 to 1.30 GPa for pure PMMA and PMMA containing 11.3 wt.% intercalated montmorillonite, respectively. This again attests for the inefficiency of intercalated structures to improve the stiffness of the so-obtained nanocomposites. Exfoliation of layered materials such as magadiite in an elastomeric epoxy matrix [74] also gives rise to a noticeable increase in the Young's modulus of the obtained composites as depicted in Fig. 33.

Fig. 33. Comparison of the evolution of tensile modulus with filler content for nanocomposites based on an epoxy matrix and various organomodified fillers. C18A-montmorilloniteˆmontmorillonite modified with octadecylammonium C18Amagadiiteˆmagadiite modified with octadecylammonium C18A1M-magadiiteˆmagadiite modified with methyloctadecylammonium (reproduced from [74] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 34. Tensile modulus vs. organoclay loadings for elastomeric polyurethane/clay nanocomposites (reproduced from [75] with permission).

In Fig. 33 is shown the evolution of the tensile modulus with filler loading for three layered materials: a montmorillonite modified with octadecylammonium cation (C18A-montmorillonite), a magadiite modified with the same alkylammonium (C18A-magadiite) and a magadiite modified with methyl-octadecylammonium cation (C18A1M-magadiite). This figure shows a much significant increase of the tensile modulus for the montmorillonite-based nanocomposites for filler contents higher than 4 wt.%. The authors explain this behavior by the difference in layer charge density for magadiite and montmorillonite. Organomagadiites have a higher layer charge density and consequently a higher alkylammonium content than organomontmorillonite. As the alkylammonium ions interact with the epoxy resin while polymerizing, dangling chains are formed. More of these chains are thus formed in presence of organomagadiites. These dangling chains are known to weaken the polymer matrix by reducing the degree of network cross-linking, then compromising the reinforcement effect of the silicate layer exfoliation. In a pure elastomeric matrix, exfoliation does not appear to be a prerequisite to improve the material stiffness. Indeed, it has been reported that cross-linked soft polyurethane-based nanocomposites can be characterized by a two-fold increase of the tensile modulus upon filling with 10 wt.% of organoclay (Fig. 34) [75]. Here, no exfoliation has been observed, rather the highly regular intercalation, with interlayer Ê (meaning that more than one polymer chain is intercalated), would explain spacing as high as 50.8 A the mechanical properties improvement. A large increase in the tensile modulus for an exfoliated structure is also observed for thermoset matrices [71,72]. Fig. 35 shows the evolution of modulus for various amine-cured epoxy-based nanocomposites filled with 2 wt.% montmorillonite previously modified by alkylammonium cations of different length [72]. While the montmorillonite modified with butylammonium only gives an intercalated structure with a low tensile modulus, the other three nanocomposites with alkyl chains of 8, 12 and 16 carbons are characterized by exfoliated structures as determined by TEM and XRD, and consecutively give much higher modulus values. However, Zilg et al. [73] have reported about rather weak stiffness improvements in the case of anhydride-cured epoxy-based nanocomposites when true exfoliated structures were observed. For these authors, the real key for the matrix stiffness improvement resides in the formation of supramolecular assemblies obtained by the presence of dispersed anisotropic laminated nanoparticles. They also describe a stiffening effect when the montmorillonite is modified by a

43

44

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 35. Dependence of tensile modulus of amine-cured epoxy/clay nanocomposites on onium ion carbon number at clay loadings of 2 wt.%.

functionalized organic cation (carboxylic acid or hydroxyl groups) that can interact with the matrix during curing. 4.1.1.2. Stress at break. In thermoplastic-based (intercalated or exfoliated) nanocomposites, the stress at break, which expresses the ultimate strength that the material can bear before break, may vary strongly depending on the nature of the interactions between the matrix and the filler as shown in Table 15. Even if caution has to be taken concerning the absolute tensile stress at break values (most of the described nanocomposites were obtained by in situ polymerization so as properties may be influenced by changes in the matrix molecular parameters as well), the differences observed are usually sufficiently sizeable to draw some conclusions. Filled polymers such as exfoliated nylon-6Table 15 Tensile stress evolution for nanocomposites based on various thermoplastic matrices Matrix

Matrix tensile stress (MPa)

Nylon-6

68.6 68.6 68.6

4.7 5.3 4.1

PMMA

53.9 53.9

12.6 20.7

PP

31.4 32.6

5.0 4.8

PS

28.7 28.7 28.7 28.7

11.3 17.2 24.6 34.1

a

Nanofiller content (wt.%)

Nanocompo site type

Nanocomposite tensile stress (MPa)

NCHa b L-NCH One-pot-NCHc

97.2 97.3 102

Intercalated Intercalated

62.0 62.0

Intercalatedd Intercalatede (‡exfoliated?) Intercalated Intercalated Intercalated Intercalated

29.5 31.7 21.7 23.4 16.6 16.0

NCH: exfoliated nanocomposite prepared by in situ intercalative polymerization of e-caprolactam in protonated aminododecanoic acid modified montmorillonite. b L-NCH: exfoliated nanocomposite prepared by in situ intercalative polymerization of e-caprolactam in montmorillonite pre-intercalated with e-caprolactam. c One-pot-NCH: exfoliated nanocomposite prepared by in situ intercalative polymerization of e-caprolactam activated by phosphoric acid in Na-montmorillonite. d PP added with PP-MA so as a PPCH 1/1 is reached (see Table 14). e PP added with PP-MA so as a PPCH 1/3 is reached (see Table 14).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

based nanocomposites prepared following different methods [18,50,51] or intercalated PMMA-based nanocomposites [43] exhibit an increase in the stress at break, that is usually explained by the presence of polar (PMMA) and even ionic interactions (nylon-6 grafted onto the layers) between the polymer and silicate layers. This increase appears to be much more pronounced in case of nylon-6 which has both an exfoliated structure and ionic bonds with the silicate layers. As far as polypropylene-based nanocomposites are concerned [88], no or only very slight tensile stress enhancement are measured (Table 15). This behavior can be partially explained by the lack of interfacial adhesion between apolar PP and polar layered silicates. Addition of maleic anhydride modified polypropylene to the polypropylene matrix has, however, proved to be favorable to the intercalation of the PP chains and maintains the ultimate stress at an acceptable level. In PSintercalated nanocomposites [45], ultimate tensile stress is even much decreased compared to PP matrix and further drops down at higher filler content. This lack of properties is attributed by the authors to the fact that only weak interactions exist at the poly(styrene)-clay interface contrary to the previous compositions in which (stronger) polar interactions may take place, strengthening the filler matrix interface. Epoxy resins-based nanocomposites display a totally different behavior depending upon their glass transition temperature, located above or below room temperature. In high Tg epoxy thermosets [72,73], neither intercalated nor exfoliated nanosilicates lead to an improvement of the tensile stress at break, they rather make the materials more brittle. This effect appears to be generally more pronounced for intercalated structures than for exfoliated ones. In contrast, nanocomposites based on both epoxy [71,74] and polyurethanes [75] elastomeric matrices exhibit a sizeable increase in tensile stress at break upon the addition of small quantities of nanofillers. This increase follows qualitatively those observed previously for the Young's modulus measurements. The same increase is also observed for silicon rubber-based nanocomposites [93] as shown in Fig. 36. In this study, the increase in tensile stress at break in function of the volume content of filler for a partially-intercalated partially-exfoliated nanocomposite is compared to a composite filled with silica anisotropic nanoparticles (5±20 nm), coined as aerosilica. It appears thus that for low glass transition temperature cured nanocomposite materials, the tensile strength increase does not rely upon the aspect ratio of the dispersed particle but rather on the presence of nanoparticles dispersed in the cross-linked soft matrix.

Fig. 36. Tensile strength of nanocomposites vs. volume content of filler: (square) silicone rubber/organo-modified montmorillonite nanocomposites; (triangle) silicone rubber/aerosilica nanocomposites (reproduced from [93] with permission).

45

46

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 37. Comparison of the strain at break values for an exfoliated epoxy/magadiite nanocomposite prepared from magadiite modified with methyl-octadecylammonium ion (C18A1M), an intercalated nanocomposite prepared from magadiite modified with trimethyloctadecylammonium ion (C18A3M) and a conventional composite prepared from magadiite modified with octadecylammonium ion (C18A) (reproduced from [74] with permission).

4.1.1.3. Elongation at break. The effect of nanocomposite formation on the elongation at break has not been widely investigated. When dispersed in thermoplastics such as for intercalated PMMA [43] and PS [45] or intercalated±exfoliated PP, the elongation at break is reduced. In the last case, the decrease is very important, dropping from 150 and 105% for a pure PP matrix and a 6.9 wt.% nonintercalated clay microcomposite, respectively, down to 7.5% in the better case for a PP-based nanocomposite filled with 5 wt.% silicate layers. Interestingly enough, such a loss in ultimate elongation does not occur in elastomeric epoxy [74] or polyol polyurethane matrices [75]. Rather, the addition of a nanoclay in cross-linked matrices triggers an increase of the elongation at break as clearly depicted in Fig. 37. When a conventional composite (magadiite exchanged with octadecylammonium cation, C18A) is prepared, a drop in the elongation at break is observed as expected for such a material while an intercalated nanocomposite (magadiite exchanged with trimethyloctadecylammonium cation, C18A3M) tends to slightly improve this property. But the exfoliated nanocomposite as prepared with methyloctadecylammonium cation (C18A1M) displays a large increase of the elongation at break. The improvement in elasticity may be attributed in part to the plasticizing effect of the gallery oniums and to their contribution to the formation of dangling chains but also probably to conformational effects at the clay-matrix interface. The combination of improved stiffness (Young's modulus), toughness (stress at break) and elasticity (strain at break) is quite exceptional and make elastomeric nanocomposites a new family of highly performant materials. Another matrix which exhibits both an increase in stress and elongation at break is poly(imide), PI [102]. Indeed, when filled by montmorillonite exchanged with hexadecylammonium cation (16CNH), these properties increase with the filler loading at least up to a 5 wt.% filler content. At higher filler content, both properties experience a sharp drop towards values lower than those recorded for the filler-free matrix (see Fig. 38). This behavior is explained by the formation of non-exfoliated aggregates at higher filler content, that makes these composites much more brittle. 4.1.2. Impact properties Impact properties have been measured for nylon-6-based nanocomposites prepared either by in situ intercalative polymerization of e-caprolactam using protonated aminododecanoic acid-

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 38. Tensile strength and elongation at break in function of filler content for poly(imide)-based nanocomposites filled with montmorillonite modified with hexadecylammonium ion (reproduced from [102] with permission).

exchanged montmorillonite [18] or by melt intercalation of nylon-6 in octadecylammoniumexchanged montmorillonite [85]. Both methods lead to exfoliated nanocomposites especially when the filler content does not exceed 10 wt.% (at higher filler level, melt-intercalation provides partiallyexfoliated±partially-intercalated materials). The formation of nylon-6-based nanocomposites does not reduce too much the impact properties, whatever the exfoliation process used. In the case of in situ intercalative polymerization, the Izod impact strength is reduced from 20.6 to 18.1 J/m when 4.7 wt.% of nanoclay is incorporated. Charpy impact testing shows similar reduction in the impact strength with a drop from 6.21 kJ/m2 for the filler-free matrix down to 6.06 kJ/m2 for the 4.7 wt.% nanocomposite. Fig. 39 shows that the decrease in the Izod impact strength of melt-intercalated nylon-6-based nanocomposites is not too much pronounced over a relatively large range of filler content. This relatively good resistance to impact, together with a high Young's modulus, good flexural modulus and a remarkable enhancement of the increase in the heat distortion temperature (i.e. a measure of the material softening point) going from 658C for pure nylon-6 to more than 1508C for nanocomposites, have allowed this material to replace glass fiber reinforced nylon or poly(propylene) in the production of timing belt covers of automotive engines [95]. The belt cover, obtained by injection-molding, shows good rigidity, excellent thermal stability and no wrap. It moreover saves weight up to 25% due to the very small content of inorganic material in the final composition.

Fig. 39. Effect of clay content on notched Izod impact strength of nylon-6/clay nanocomposites obtained trough melt intercalation (reproduced from [85] with permission).

47

48

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 40. Temperature dependence of E0 and tan d for unfilled poly(styrene) and an intercalated nanocomposite (17.2 wt.% of montmorillonite) (reproduced from [45] with permission).

4.1.3. Dynamic mechanical analysis Dynamic mechanical analysis (DMA) measures the response of a given material to a cyclic deformation (usually tension or three-point flexion type deformation) as a function of the temperature. DMA results are expressed by three main parameters: (i) the storage modulus (E0 ), corresponding to the elastic response to the deformation; (ii) the loss modulus (E00 ), corresponding to the plastic response to the deformation and (iii) tan d, that is the (E0 /E00 ) ratio, useful for determining the occurrence of molecular mobility transitions such as the glass transition temperature. DMA analysis has been studied to track the temperature dependence of storage modulus upon the formation of an intercalated PS nanocomposites [45]. Fig. 40 shows the temperature dependence of E0 and tan d for pure PS and a nanocomposite intercalated with 17.2 wt.% of Na-montmorillonite as synthesized by exfoliation±adsorption during emulsion polymerization (see Section 3.1.3). No significant difference in E0 can be seen in the investigated temperature range, indicating that intercalated nanocomposites do not strongly influence the elastic properties of the matrix. On the other hand, the shift and broadening of the tan d peak towards higher temperatures for the nanocomposite indicate an increase in the glass transition temperature together with some broadening of this transition. This behavior has been ascribed to the restricted segmental motions at the organic±inorganic interface neighborhood of intercalated compositions. Intercalation of PS sequences plays nevertheless a much more important role in the dynamic mechanical properties of symmetric (styrene±butadiene±styrene) block copolymers [91]. As previously described (see Section 3.3.1) when SBS are melt blended with a montmorillonite modified by dimethyldioctadecylammonium cation, a nanocomposite is formed, in which only the PS blocks can intercalate within the layered silicates. This particular structure provides a material with a sizeable improvement of the storage modulus at 258C as depicted in Fig. 41. This figure compares the values of the storage modulus for two sets of samples for which the filler content is varied from 0 to 30 wt.%. The first series (open squares) displays the values recorded for nanocomposites filled with organo-modified clay and the second one (filled squares) shows the results obtained for composites prepared by melt-blending the SBS matrix and Na-montmorillonite under the same conditions (microcomposites). One can clearly see the large increase in elastic modulus for nanocomposites while microcomposites do not present any improvement for this property, whatever the filler content be. Increase in storage modulus related to a better nanoclay dispersion is further demonstrated in the case of polypropylene-based nanocomposites (see Table 14). The degree of nanofiller dispersion

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 41. Trend of the storage modulus E0 at 258C for SBS-based nanocomposites (&&) and microcomposites (&&) as a function of the filler level (reproduced from [91] with permission).

can be tuned up either by using different amounts of a given PP modified with maleic anhydride (acid valueˆ52 mg KOH/g) [88] or can even be drawn towards the formation of exfoliated nanocomposites by playing with the relative functionalization content all along of the modified PP (using a PP-MA with an acid valueˆ26 mg KOH/g) [87]. Varying the relative amount of PP-MA within the PP matrix highly modifies the temperature dependence of the storage modulus reduced to the value of unfilled PP matrix as described in Fig. 42a). Below Tg (located around 138C for all the composites) the relative modulus values of the nanocomposites are not so much enhanced (1.2±1.3) compared to the pure matrix while above Tg, an important increase of the moduli is observed, reaching a maximal value of 1.76 around 808C, in the case of the PPCH 1/3 composite. On the other hand, the relative storage modulus of the microcomposite (PPCC) stays relatively low and quickly reaches a plateau value around 1.26. One can further observe that the storage moduli show a sharp decline above 1408C. The main reason for this decline resides in the fact that the softening point of the PP-MA matrix is reached at this temperature, strongly reducing the elastic response of the material. Fig. 42b shows the relative storage modulus values in reference to the unfilled PP/PP-MA blends properties instead of the pure PP matrix. The very large increase observed for the PPCH 1/3 composition can be directly related to the far better dispersion obtained for this nanocomposite (see Section 3.3.1). When the dispersion is further improved by the use of a more compatible PP-MA (code 1001; 26 mg KOH/g), higher storage moduli are reached (Fig. 43). A two-fold increase in the relative moduli (E0 /E0 -matrix) is even measured for a nanocomposite based on organo-modified montmorillonite nanoparticles (PPCH-C18-Mt/1001). Interestingly enough, a maximum value of 2.4 is reached by substituting synthetic fluorinated mica (PPCHC18-Mc/1001) for previously studied montmorillonite. The behavior difference for the two types of filler was not explained by the authors but it is more likely related to the respective aspect ratio of the dispersed particles since mica layers are known to be much longer than montmorillonite layers. The influence of dispersion and length of the layered particles is further demonstrated in case of poly(imide)-based nanocomposites using various organoclays (hectorite, saponite, montmorillonite and synthetic mica, see Table 3) [38]. In this study, exfoliated structures were obtained for mica and montmorillonite clays while a partially-exfoliated±partially-intercalated structure was found for saponite and a mainly intercalated morphology was attributed to the hectorite-based nanocomposite. Fig. 44 shows the temperature dependence of the storage modulus for these nanocomposites filled with 2 wt.% of clay and for the unfilled matrix.

49

50

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 42. Temperature dependence of the relative storage modulus (E0 ): (a) storage modulus of PPCHs nanocomposites and PPCC microcomposite relative to the storage modulus of the PP matrix and (b) storage modulus of PPCH 1/1 and 1/3 relative to the corresponding PP-MA matrices. See Table 14 for code explanation (reproduced from [88] with permission).

At a given temperature, higher storage moduli results from the better nanofiller dispersion. The huge difference between exfoliated montmorillonite and exfoliated mica-based nanocomposites may be again explained by the respective aspect ratio of the dispersed silicate layers, with lengths of 0.218 and 1.23 mm, respectively, for montmorillonite and synthetic mica, as observed by TEM. Finally, DMA studies carried out on organoclays exfoliated within cross-linked matrices reveals again a very marked improvement of the storage modulus, especially above Tg. For instance, for epoxy matrix below Tg [69], a 58% increase in modulus results from the dispersion of 4 vol.% montmorillonite with the formation of a well-ordered exfoliated nanocomposite (silicate layers Ê ). At 408C, E0 equals 2.44 and 1.55 GPa for the nanocomposite separated by approximately 100 A and the unfilled cross-linked matrix, respectively. But above Tg, e.g. at 1508C, the storage modulus improvement reaches a 4.5 factor with E0 values of 11 and 50 MPa for the unfilled and filled epoxy, respectively. Similar behavior is observed for room temperature elastomer such as nitrile rubber [95], A three-fold increase of the storage elastic modulus is noted by the simple dispersion/exfoliation of 10 parts of organoclay per 100 parts of rubber, with a modulus as high as 8.8 MPa. This value corresponds to what can be obtained with the same matrix filled with 40 parts of carbon black per 100 parts of rubber, thus reducing by a factor of four the amount of filler.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 43. Relative dynamic storage moduli (E0 /E0 -matrix) of the PP-based nanocomposites based on maleic anhydride modified PP (PP-MA/1001) to that of corresponding PP/1001 blends taken as the matrix in function of the temperature (reproduced from [87] with permission).

In summary, the storage elastic modulus appears to be substantially enhanced at temperatures above Tg for exfoliated nanocomposites filled with layered silicates of high aspect ratio. A possible explanation for such an improvement could be the creation of a three-dimensional network of interconnected long silicate layers, strengthening the material through mechanical percolation. 4.2. Thermal stability and flame retardant properties Another highly interesting property exhibited by polymer-layered silicate nanocomposites concerns their increased thermal stability but also their unique ability to promote flame retardancy at quite low filling level through the formation of insulating and incombustible char. 4.2.1. Thermal stability The thermal stability of a material is usually assessed by thermogravimetric analysis (TGA) where the sample mass loss due to volatilization of degraded by-products is monitored in function of a temperature ramp. When the heating is operated under an inert gas flow (nitrogen, helium, . . .), a non-oxidative degradation occurs while the use of air or oxygen allows to follow the oxidative degradation of the sample. The first indication of thermal stability improvement in nanocomposites appears in the seminal work by Blumstein [103] who studied the thermal stability of PMMA intercalated within

Fig. 44. Temperature dependence on storage elastic modulus for poly(imide)-based nanocomposites filled with 2 wt.% of organo modified synthetic mica, montmorillonite, saponite and hectorite (reproduced from [39] with permission).

51

52

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

montmorillonite. In this work, the author shows that a 10 wt.% clay intercalated PMMA (produced by free radical polymerization of the intercalated monomer) degrades at temperature 40±508C superior to the degradation of the pure unfilled PMMA matrix. He also found that the thermal stability of the PMMA extracted from the montmorillonite was higher than for a PMMA conventionally produced in solution. The higher stability of PMMA synthesized by in situ intercalative polymerization is more likely due to a decrease in the relative amount of PMMA endcapped by carbon±carbon double bond, as a result of reduced propensity to disproportionation reactions. Extracted PMMA chains were nevertheless less stable than when intercalated, and the author proposed that the enhanced thermal stability of the PMMA-based nanocomposites was not only due to difference in chemical structure, but also to restricted thermal motion of the macromolecule in the silicate interlayer. Since then and more particularly in the 1990s, several authors have drawn attention to the thermal stabilization brought by the nanocomposites. Burnside and Giannelis [92] have measured by TGA (under nitrogen flow) the thermal stability of cross-linked poly(dimethylsiloxane) in which 10 wt.% of organomontorillonite was exfoliated. When compared to unfilled cross-linked PDMS (Fig. 45), the nanocomposite TGA trace shows a drastic shift of the weight loss towards higher temperature, with a stabilization as high as 1408C at 50% weight loss. The authors attributed the much better thermal stability to hindered out-diffusion of the volatile decomposition products (mainly cyclic silicates), as a direct result of the decrease in permeability, usually observed in exfoliated nanocomposites (see Section 4.3). More recently, another group [93] has produced nanocomposites based on cross-linked PDMS using slightly different operating conditions in order to produce mainly intercalated structures. In this case, the increase in thermal stability for a nanocomposite intercalated with 8.1 vol.% of organomontmorillonite was limited to about 608C at 50% of weight loss. The authors proposed other possible origins for the observed thermal stability improvement (which was also reported, in the same study, for silica-based nanocomposites) such as some inactivation of the centers active in silicone main chain decomposition by interaction with the filler or by prevention of the unzipping degradation from occurring through physical and chemical cross-linking points built up between polymer chains and filler particles. Increase in thermal stability has also been reported for intercalated nanocomposites prepared by emulsion polymerization of methyl methacrylate [43], styrene [45] and epoxy precursors [44] in the

Fig. 45. TGA traces for PDMS (solid line) and PDMS nanocomposite (dashed line) containing 10 wt.% organo-modified montmorillonite (reproduced from [92] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 46. Decomposition onset temperature as a function of filler loading of PS-based nanocomposites obtained with montmorillonite modified with: (filled square) dimethyl(hydrogenated tallow alkyl)benzyl ammonium ion (filled circle) dimethyl di(hydrogenated tallow alkyl) ammonium ion (filled triangle) dimethyl(hydrogenated tallow alkyl) 2-ethylhexyl ammonium ion (open square) Na-montmorillonite (reproduced from [60] with permission).

presence of water swollen Na-montmorillonite. In every case, a high temperature increase in the decomposition onset was observed. Doh and Cho [60] have measured by TGA under nitrogen atmosphere the onset of thermal decomposition of intercalated PS-based nanocomposites produced by in situ polymerization of styrene within various organo-modified montmorillonites (see Table 8). In Fig. 46 are collected these decomposition onset temperatures of PS-based nanocomposites filled with increasing filler content together with a Na-montmorillonite-based microcomposite for the sake of comparison. It is clearly seen that a large increase in the onset of decomposition occurs for nanocomposites at very low filler content and quickly levels off. The threshold is reached for a filler content as low as 0.3 wt.% when intercalating an organoclay (modified with a dimethylbenzyloctadecylammonium cation) well compatible with PS. In contrast, Na-montmorillonite does not modify a lot the decomposition onset of the PS matrix. This is another widely searched characteristic feature of nanocomposites in which the thermal properties improvement arises at very low filler content, often making the obtained material cheaper, lighter and easier to process than more conventional microcomposites. Another key factor that may determine the extent of the thermal stabilization in nanocomposites could also arise from the actual nature of the thermal degradation mechanism, often different from a polymer to another. For example, when poly(imide) exfoliated nanocomposites [41] are thermally degraded under nitrogen, their thermal stability is only enhanced by about 258C (at 50% of weight loss) which is much less than the 1408C jump observed in exfoliated PDMS nanocomposites. Without any doubt, the chemical nature of the studied polymeric material and its degradation mechanism play here an important role. Finally, the experimental conditions of the material degradation have proved to highly influence the history and mechanism of the degradation as well. The thermal stability of EVA partiallyintercalated±partially-exfoliated nanocomposites have been investigated through TGA under helium (thermodegradation) and under air flow (thermooxidative degradation) [90]. EVA is known to degrade in two consecutive steps, the first one, identical for both oxidative and non-oxidative degradations, consists in the loss of acetic acid and occurs between 350 and 4008C. The second step involves the thermal degradation of the so obtained unsaturated backbone either by radical scission (non-oxidative route) or by thermal combustion (oxidative route). Fig. 47a presents the TGA traces

53

54

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 47. Influence of the purge gas on the TGA and DTG traces EVA-based nanocomposite (5 wt.% of organo-modified montmorillonite, long dashes), microcomposite (5 wt.% of Na-montmorillonite small dashes) and the unfilled matrix (solid line): (a) under helium flow and (b) under air flow.

and their derivatives (DTG) for, respectively, the unfilled EVA matrix with a vinyl acetate content of 27 wt.%, a 5 wt.% Na-montmorillonite/EVA microcomposite and a 5 wt.% modified montmorillonite/EVA nanocomposite as measured under helium at 208C/min. Fig. 47b presents the same TGA and DTG traces but recorded under air flow.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Table 16 Maximal temperature at the main degradation peak (DTG) as measured under air flow at 208C/min for EVA and EVAbased nanocomposites with different nanoclay contents Filler content (wt.%)

Degradation temperature maximum (8C)

0 1 2.5 5 10 15

452.0 453.4 489.2 493.5 472.0 454.0

One can note first that in both cases, the Na-montmorillonite (microcomposite) does not influence the thermal degradation of the matrix. The most striking observation comes out by comparing the second degradation step under helium and under air flow. While, under helium, the nanocomposite experienced only a very slight loss in thermal stability (48C), a sizeable increase of more than 408C at the maximum of the DTG curves is measured under air. An explanation for this behavior may be found in the formation of char that clearly appears under oxidative degradation. Char may act as a physical barrier between the polymer medium and the superficial zone where flame combustion occurs. Comparison of different nanofiller contents in EVA-based nanocomposites on the degradation peak in the DTG curve under oxidative degradation, has been also very informative. Results are collected in Table 16. Optimal thermal stabilization is obtained at a filler content of ca. 2.5±5 wt.%. Below this value no thermal stabilization is observed while increasing too much the amount of nanofiller also decreases the thermal stability. Such a behavior could account from the change in relative proportion of exfoliated and intercalated species with the filler content. At low filler content, exfoliation dominates but the amount of exfoliated particles is not high enough to promote the thermal stability through char formation. When increasing the filler content, relatively more exfoliated particles are formed, char forms more easily and increases the thermal stability of the nanocomposite until 2.5± 5 wt.% of nanofiller is reached. At higher levels, equilibrium between exfoliation and intercalation is drawn towards intercalation and, even if char is still formed in high quantity, the morphology of the nanocomposite does not allow for maintaining a good thermal stability. This explanation, likely valid for EVA nanocomposites may not be applicable to other polymer matrices as demonstrated in a study on poly(etherimide) nanocomposites [104] where intercalated nanolayers were found to be better thermal stabilizers than the exfoliated ones. 4.2.2. Flame retardancy The flame retardant properties of nanocomposites have been very recently reviewed in detail by Gilman [105]. The main bench-scale method used to measure important parameters in the flame retardant behavior of a material (heat release rate, peak of heat release rate, heat of combustion, . . .) is Cone calorimetry. In a typical experiment, the sample is exposed to a given heat flux (often taken as 35 kW/m2) and the heat release rate (HRR) as well as the mass loss rate are recorded as a function of time. It is worth noting that reduction of the peak HRR is the most clear-cut evidence for the efficiency of a flame retardant. Moreover, gas and soot production are also measured. As a typical example, Fig. 48 shows the HRR plot obtained for nylon-6 and a nylon-6 exfoliated nanocomposite (5 wt.% of exfoliated montmorillonite). A 63% reduction in the peak HRR is clearly observed for the nanocomposite. Cone calorimetry experiments have been carried out on other nanocomposites such as exfoliated nylon-12 (2 wt.%

55

56

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 48. Comparison of the heat release rate (HRR) plots for nylon-6 and nylon-6 layered silicate nanocomposite (5 wt.%) at 35 kW/m2 heat flux, showing a 63 wt.% reduction in HRR for the nanocomposite (reproduced from [105] with permission).

organoclay), exfoliated poly(methylmethacrylate-co-dodecylmethacrylate) [106], intercalated PS (3 wt.%) or intercalated PP (2 wt.%) and for each material, a significant decrease in the peak HRR is recorded while the heat of combustion, smoke and the carbon monoxide yields (other important properties in flammability concern) are usually not increased. These data tend to demonstrate that the improvement in flame retardancy does not occur by a process in the gas phase but rather by a modification of the combustion process in the condensed phase. Experiments carried out in a radiative gasification apparatus [107] have allowed to determine that the flame retardant effect of nanocomposites mainly arises from the formation of char layers obtained through the collapse of the exfoliated and/or intercalated structures. This multilayered silicate structure may act as an excellent insulator and mass transport barrier, slowing down the escape of the volatile decomposition products as observed in nylon-6 but also in thermoset nanocomposites [108]. Whatever the nature of the matrix (thermoplastics or thermosets) and whatever the structure of the nanocomposite (exfoliated or Ê ) was found for the recovered chars as intercalated), always the same interlayer spacing (13 A analyzed by XRD, implying the formation by combustion of a residue of the same nature. Nylon-6 nanocomposite filled with 2 wt.% nanoclay has also been used as an additive to replace pentaerythritol (in order to avoid exudation and water solubility) in an intumescent flame retardant formulation using ammonium polyphosphate, APP [109]. This new formulation has shown very good fire retardant properties, increasing the low oxygen index (LOI) values by 5% for the best APP/ nylon-6 nanocomposite composition and highly decreasing the HRR values. 4.3. Gas barrier properties The high aspect ratio characteristic of silicate nanolayers in exfoliated nanocomposites has been found to highly reduce the gas permeability in films prepared from such nanomaterials. The permeability to carbon dioxide has been measured for the partially-exfoliated poly(imide)based nanocomposites prepared by Lan et al. [40]. Interestingly, the relative permeability values, i.e. Pc/Pp where Pc and Pp stand for the composite and the unfilled polymer permeability, respectively, have been plotted in function of the filler volume fraction (Fig. 49). The curve fitting has been achieved by using a theoretical expression allowing the prediction of gas permeability in function of the length and width of the filler particles as well as their volume

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Fig. 49. CO2 permeability of polyimide clay composites prepared by curing CH3(CH2)17NH3‡ montmorillonite-poly(amic acid) films at 3008C. Curves A and C are calculated for filler with an aspect ratio of 20 and 2000, respectively, Curve B was generated by least squares fitting of the permeability equation to the experimental data. The inset illustrates a possible selfsimilar aggregation mechanism for the clay plates (reproduced from [40] with permission).

fraction within the PI matrix. The best fitting is actually obtained for an aspect ratio of 192. Surprisingly, this value is much smaller than what could be awaited for a truly exfoliated system structure, meaning an aspect ratio of ca. 2000 for montmorillonite single layers [40]. Explication for such a low aspect ratio is illustrated by the inset shown in Fig. 49, where can be seen a sketch of the proposed nanocomposite structure. It consists of the so called `self-similar clay aggregation mechanism' in which face±face associated and elongated layers are skipped in staircase-like fashion. These self-similar structures can, therefore, exhibit enhanced aspect ratio. The effect on water permeability of both partially and totally exfoliated poly(imide)-based nanocomposites has been reported by Yano et al. [39], using organoclay with different layer lengths (see Table 3). Fig. 50 presents the clay length dependence of the relative permeability coefficient for poly(imide) filled with 2 wt.% of organoclay, either exfoliated montmorillonite and synthetic mica, or intercalated clay tactoids (hectorite and saponite). A relatively good agreement between the experimental values and the corresponding theoretical curve can be achieved, as the length of the clay increases, the relative permeability decreases drastically. In other words, it does mean that the best gas barrier properties will be obtained by fully exfoliated rather long layered silicates.

Fig. 50. Clay length dependence on the relative permeability coefficient for poly(imide)/clay nanocomposites (reproduced from [39] with permission).

57

58

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

Permeability to water vapor has also been investigated for exfoliated nanocomposite based on poly(e-caprolactone) (PCL), synthesized by in situ intercalative polymerization of the lactone monomer inside organo-modified montmorillonite [53]. In order to produce nanocomposite films with different filler contents, PCL nanocomposite containing 15 wt.% (6 vol.%) of exfoliated montmorillonite was separately prepared and mixed in variable composition with a commercial preformed PCL by co-dissolution in toluene followed by solvent casting. Again, a dramatic drop in the relative permeability of the polymer coefficient is triggered by dispersing increasing amounts of layered nanofiller. A relative permeability of 0.2 (unfilled PCL permeabilityˆ1) is measured for a filler loading of 4.8 vol.%. Fitting of the Pc/Pp versus filler volume fraction dependency lead the authors to conclude to an aspect ratio of 70, again well below the value expected for exfoliated montmorillonite. The authors explained this apparent discrepancy by the fact that the fitting model was originally developed for particles totally oriented parallel to the film plane. However, under the experimental conditions used for the film formation, the silicate layers could be not so well aligned flat along the film surface, with even possible filler aggregation. Finally, it is worth pointing out that Bayer has recently commercialized a new grade of plastic films for food packaging, Durethan1 LPDU 601, based on nylon-6 exfoliated nanocomposites with improved gas barrier properties (oxygen transmission rate divided by a factor of two compared with the pure nylon-6), improved transparency and gloss and increased tensile modulus [110]. 4.4. Miscellaneous 4.4.1. Ionic conductivity Nanocomposites have been also considered by Vaia et al. [96] to tune ionic conductivity of PEO. An intercalated nanocomposite obtained by melt intercalation of poly(ethylene oxide) (40 wt.%) into Li-montmorillonite (60 wt.%) has shown to enhance the stability of the ionic conductivity at lower temperature (see Fig. 51) when compared to more conventional PEO/LiBF4 mixture. This improvement is explained by the fact that PEO is not able to crystallize when intercalated, hence eliminating the presence of crystallites, non-conductive in nature. The conductivity of PEO/ Li-montmorillonite nanocomposite is 1.610ÿ6 S/cm at 308C and exhibits a weak temperature dependence with an activation energy of 2.8 kcal/mol. The higher ionic conductivity at ambient

Fig. 51. Arrhenius plots of ionic conductivity for LiBF4/PEO and PEO Li‡ montmorillonite intercalated nanocomposite (reproduced from [96] with permission).

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

temperature compared to conventional LiBF4/PEO electrolytes combined with a single ionic conductor character makes those nanocomposites new promising electrolyte materials. 4.4.2. Thermal expansion coefficient Due to the high aspect ratio of the exfoliated silicate layers, the thermal expansion coefficient of poly(imide)-based nanocomposites prepared by Yang et al. [102] with hexadecylammonium cation exchange montmorillonite can be strongly reduced, going from 3.610ÿ5 Kÿ1 to values as low as 1.5510ÿ5 Kÿ1 when 10 wt.% of nanofiller is dispersed. A noticeable decrease of 45% (1.9610ÿ5 Kÿ1) was already observed with only 1 wt.% of nanoclay. 4.4.3. Other properties Finally, nanocomposites have been used in highly technical domains such as the improvement of ablative properties in aeronautics [111], the potentiality to use polyaniline-based nanocomposite as electrorheological sensitive additive [112] or the combination of dispersed layered silicates in a liquid crystal medium for the production of stable electro-optical devices exhibiting a bistable and reversible electro-optical effect between a light scattering opaque state and a transparent state [113]. 5. Conclusions The large array of improved thermo-mechanical properties attained at very filler content (5 wt.% or less) together with the ease of production through simple processes such as melt intercalation, directly applicable by extrusion or injection molding make layered silicate-based nanocomposites a very promising new class of materials. They are already commercially available and applied in car and food packaging industries. Undoubtedly, the unique combination of their key properties and potentially low production costs paves the way to much broader range of applications. Furthermore, the quite low filler level required to display sizeable properties enhancement makes them competitive with other materials. Their incineration produces ceramic chars in low yield and the very limited filler content makes them compatible with recycling process. Nevertheless, a lot of research remains to be carried out in order to fully understand factors such as exfoliation versus intercalation driving forces yielding the different nanocomposite structures. Moreover, a much better understanding of the actual structure/properties relationships has to be fulfilled in some important area such as fire retardancy or physico-mechanical properties. Finally, it is worth pointing out that new types of layered nanoparticles have recently been reported and their ability to form nanocomposites with enhanced properties has been proposed. For example, superconductive nanofillers [114] and magnetic particles [115] have shown to be capable of exfoliation under controlled conditions and one can foresee at short-term the development of nanocomposite materials articulated onto such multifunctional nanofillers. Acknowledgements Financial support from `ReÂgion Wallonne' and European Community (FEDER and FSE) in the frame of `PoÃle d'Excellence Materia Nova: Objectif 1' is greatly acknowledged. References [1] J.E. Mark, Ceramic reinforced polymers and polymer-modified ceramics, Polym. Eng. Sci. 36 (1996) 2905±2920. [2] E. Reynaud, C. Gauthier, J. Perez, Nanophases in polymers, Rev. Metall./Cah. Inf. Tech. 96 (1999) 169±176.

59

60

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

[3] T. von Werne, T.E. Patten, Preparation of structurally well defined polymer±nanoparticle hybrids with controlled/ living radical polymerization, J. Am. Chem. Soc. 121 (1999) 7409±7410. [4] N. Herron, D.L. Thorn, Nanoparticles. Uses and relationships to molecular clusters, Adv. Mater. 10 (1998) 1173±1184. [5] P. Calvert, Potential applications of nanotubes, in: T.W. Ebbesen (Ed.), Carbon Nanotubes, CRC Press, Boca Raton, FL, 1997, pp. 277±292. [6] V. Favier, G.R. Canova, S.C. Shrivastava, J.Y. Cavaille, Mechanical percolation in cellulose whiskers nanocomposites, Polym. Eng. Sci. 37 (1997) 1732±1739. [7] L. Chazeau, J.Y. Cavaille, G. Canova, R. Dendievel, B. Boutherin, Viscoelastic properties of plasticized PVC reinforced with cellulose whiskers, J. Appl. Polym. Sci. 71 (1999) 1797±1808. [8] H. Shioyama, Polymerization of isoprene and styrene in the interlayer spacing of graphite, Carbon 35 (1997) 1664±1665. [9] L. Hernan, J. Morales, J. Santos, Synthesis and characterization of poly(ethylene oxide) nanocomposites of misfit layer chalcogenides, J. Solid State Chem. 141 (1998) 327±329. [10] D.J. Harris, T.J. Bonagamba, K. Schmidt-Rohr, Conformation of poly(ethylene oxide) intercalated in clay and MoS2 studied by two-dimensional double-quantum NMR, Macromolecules 32 (1999) 6718±6724. [11] Y. Matsuo, K. Tahara, Y. Sugie, Synthesis of poly(ethylene oxide)-intercalated graphite oxide, Carbon 34 (1996) 672±674. [12] Y. Matsuo, K. Tahara, Y. Sugie, Structure and thermal properties of poly(ethylene oxide)-intercalated graphite oxide, Carbon 35 (1997) 113±120. [13] Y. Ding, D.J. Jones, P. Maireles-Torres, Two-dimensional nanocomposites: alternating inorganic±organic polymer layers in zirconium phosphate, Chem. Mater. 7 (1995) 562±571. [14] O.C. Wilson Jr., T. Olorunyolemi, A. Jaworski, L. Borum, D. Young, A. Siriwat, E. Dickens, C. Oriakhi, M. Lerner, Surface and interfacial properties of polymer-intercalated layered double hydroxide nanocomposites, Appl. Clay Sci. 15 (1999) 265±279. [15] C.O. Oriakhi, I.V. Farr, M.M. Lerner, Thermal characterization of poly(styrene sulfonate)/layered double hydroxide nanocomposites, Clays and Clay Minerals 45 (1997) 194±202. [16] B.K.G. Theng, The Chemistry of Clay-Organic Reactions, Wiley, New York, 1974. [17] M. Ogawa, K. Kuroda, Preparation of inorganic±organic nanocomposites through intercalation of organoammonium ions into layered silicates, Bull. Chem. Soc. Jpn. 70 (1997) 2593±2618. [18] Y. Kojima, A. Usuki, M. Kawasumi, A. Okada, Y. Fukushima, T. Karauchi, O. Kamigaito, Mechanical properties of nylon-6±clay hybrid, J. Mater. Res. 6 (1993) 1185±1189. [19] E.P. Giannelis, R. Krishnamoorti, E. Manias, Polymer±silica nanocomposites: model systems for confined polymers and polymer brushes, Adv. Polym. Sci. 118 (1999) 108±147. [20] G. Lagaly, Interaction of alkylamines with different types of layered compounds, Solid State Ionics 22 (1986) 43±51. [21] R.A. Vaia, R.K. Teukolsky, E.P. Giannelis, Interlayer structure and molecular environment of alkylammonium layered silicates, Chem. Mater. 6 (1994) 1017±1022. [22] E. Hackett, E. Manias, E.P. Giannelis, Molecular dynamics simulations of organically modified layered silicates, J. Chem. Phys. 108 (1998) 7410±7415. [23] C. Oriakhi, Nano sandwiches, Chem. Br. 34 (1998) 59±62. [24] M. Lerner, C. Oriakhi, in: A. Goldstein (Ed.), Handbook of Nanophase Materials, Marcel Dekker, New York, 1997, p. 199. [25] G. Lagaly, Introduction: from clay mineral±polymer interactions to clay mineral±polymer nanocomposites, Appl. Clay Sci. 15 (1999) 1±9. [26] D.J. Greenland, Adsorption of polyvinylalcohols by montmorillonite, J. Colloid Sci 18 (1963) 647±664. [27] N. Ogata, S. Kawakage, T. Ogihara, Poly(vinyl alcohol)±clay and poly(ethylene oxide)±clay blend prepared using water as solvent, J. Appl. Polym. Sci. 66 (1997) 573±581. [28] R.L. Parfitt, D.J. Greenland, Adsorption of poly(ethylene glycols) on montmorillonites, Clay Mineral 8 (1970) 305±323. [29] X. Zhao, K. Urano, S. Ogasawara, Adsorption of polyethylene glycol from aqueous solutions on montmorillonite clays, Colloid Polym. Sci 267 (1989) 899±906. [30] E. Ruiz-Hitzky, P. Aranda, B. Casal, J.C. GalvaÂn, Nanocomposite materials with controlled ion mobility, Adv. Mater. 7 (1995). [31] J. Billingham, C. Breen, J. Yarwood, Adsorption of polyamine, polyacrylic acid and polyethylene glycol on montmorillonite: an in situ study using ATR-FTIR, Vibr. Spectrosc. 14 (1997) 19±34. [32] R. Levy, C.W. Francis, Interlayer adsorption of polyvinylpyrrolidone on montmorillonite, J. Colloid Interface Sci. 50 (1975) 442±450. [33] G. Lagaly, Smectic clays as ionic macromolecules, in: A.D. Wilson, H.J. Prosser (Eds.), Development in Ionic Polymers, Elsevier, London, 1986, pp. 77±140. [34] J. Wu, M.M. Lerner, Structural, thermal, and electrical characterization of layered nanocomposites derived from sodium-montmorillonite and polyethers, Chem. Mater. 5 (1993) 835±838.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

[35] H.G. Jeon, H.-T. Jung, S.W. Lee, S.D. Hudson, Morphology of polymer silicate nanocomposites. High density polyethylene and a nitrile, Polym. Bull. 41 (1998) 107±113. [36] N. Ogata, G. Jimenez, H. Kawai, T. Ogihara, Structure and thermal/mechanical properties of poly(L-lactide)±clay blend, J. Polym. Sci.: Part B: Polym. Phys. 35 (1997) 389±396. [37] G. Jimenez, N. Ogata, H. Kawai, T. Ogihara, Structure and thermal/mechanical properties of poly(e-caprolactone)± clay blend, J. Appl. Polym. Sci 64 (1997) 2211±2220. [38] K. Yano, A. Usuki, A. Okada, T. Kurauchi, O. Kamigaito, Synthesis and properties of polyimide±clay hybrid, J. Polym. Sci.: Part A: Polym. Chem. 31 (1993) 2493±2498. [39] K. Yano, A. Usuki, A. Okada, Synthesis and properties of polyimide±clay hybrid films, J. Polym. Sci. A: Polym. Chem. 35 (1997) 2289±2294. [40] T. Lan, P.D. Kaviratna, T.J. Pinnavaia, On the nature of polyimide±clay hybrid composites, Chem. Mater. 6 (1994) 573±575. [41] H.-L. Tyan, Y.-C. Liu, K.-H. Wei, Enhancement of imidization of poly(amic acid) through forming poly(amic acid)/ organoclay nanocomposites, Polymer 40 (1999) 4877±4886. [42] C.O. Oriakhi, X. Zhang, M.M. Lerner, Synthesis and luminescence properties of a poly(p-phenylenevinylene)/ montmorillonite layered nanocomposite, Appl. Clay Sci. 15 (1999) 109±118. [43] D.C. Lee, L.W. Jang, Preparation and characterization of PMMA±clay hybrid composite by emulsion polymerization, J. Appl. Polym. Sci. 61 (1996) 1117±1122. [44] D.C. Lee, L.W. Jang, Characterization of epoxy±clay hybrid composite prepared by emulsion polymerization, J. Appl. Polym. Sci. 68 (1998) 1997±2005. [45] M.W. Noh, D.C. Lee, Synthesis and characterization of PS±clay nanocomposite by emulsion polymerization, Polym. Bull. 42 (1999) 619±626. [46] M.P. Eastman, E. Bain, T.L. Porter, K. Manygoats, R. Whitehorse, R.A. Parnell, M.E. Hagerman, The formation of poly(methyl-methacrylate) on transition metal-exchanged hectorite, Appl. Clay Sci. 15 (1999) 173±185. [47] Y. Fukushima, A. Okada, M. Kawasumi, T. Kurauchi, O. Kamigaito, Swelling behavior of montmorillonite by poly-6-amide, Clay Mineral, 23 (1988) 27±34. [48] A. Usuki, Y. Kojima, M. Kawasumi, A. Okada, Y. Fukushima, T. Kurauchi, O. Kamigaito, Synthesis of nylon-6± clay hybrid, J. Mater. Res. 8 (1993) 1179±1183. [49] A. Usuki, M. Kawasumi, Y. Kojima, A. Okada, T. Krauchi, O. Kamigaito, Swelling behavior of montmorillonite cation exchanged for o-amino acid by e-caprolactam, J. Mater. Res. 8 (1993) 1174±1178. [50] Y. Kojima, A. Usuki, M. Kawasumi, A. Okada, T. Kurauchi, O. Kamigaito, Synthesis of nylon-6-clay hybrid by montmorillonite intercalated with e-caprolactam, J. Polym. Sci. Part A: Polym. Chem. 31 (1993) 983±986. [51] Y. Kojima, A. Usuki, M. Kawasumi, A. Okada, T. Kurauchi, O. Kamigaito, One-pot synthesis of nylon-6±clay hybrid, J. Polym. Sci Part A: Polym. Chem. 31 (1993) 1755±1758. [52] P. Reichert, J. Kressler, R. Thomann, R. MuÈlhaupt, G. StoÈppelmann, Nanocomposites based on a synthetic layer silicate and polyamide-12, Acta Polym. 49 (1998) 116±123. [53] P.B. Messersmith, E.P. Giannelis, Synthesis and barrier properties of poly(e-caprolactone)-layered silicate nanocomposites, J. Polym. Sci.: Part A Polym. Chem. 33 (1995) 1047±1057. [54] D. Mecerreyes, R. JeÂroÃme, P. Dubois, Novel macromolecular architectures based on aliphatic polyesters: relevance of the coordination-insertion ring opening polymerization, Adv. Polym. Sci. 147 (1999) 1±59. [55] A. LoÈfgren, A.-C. Albertsson, P. Dubois, R. JeÂroÃme, Recent advances in ring-opening polymerization of lactones and related compounds, J. Macromol. Sci.-Rev. Macromol. Chem. Phys. C35 (1995) 379±418. [56] B.A. Rozenberg, Pure Appl. Chem. 53 (1981) 1715±1722. [57] T.K. Chen, Y.I. Tien, K.H. Wei, Synthesis and characterization of novel segmented polyurethane/clay nanocomposite via poly(e-caprolactone)/clay, J. Polym. Sci.: Part A Polym. Chem. 37 (1999) 2225±2233. [58] P.B. Messersmith, E.P. Giannelis, Polymer-layered silicate nanocomposites: in situ intercalative polymerization of e-caprolactone in layered silicates, Chem. Mater. 5 (1993) 1064±1066. [59] A. Akelah, A. Moet, Polymer±clay nanocomposites: free-radical grafting of polystyrene on to organophilic montmorillonite interlayers, J. Mater. Sci. 31 (1996) 3589±3596. [60] J.G. Doh, I. Cho, Synthesis and properties of polystyrene-organoammonium montmorillonite hybrid, Polym. Bull. 41 (1998) 511±517. [61] M.W. Weimer, H. Chen, E.P. Giannelis, D.Y. Sogah, Direct synthesis of dispersed nanocomposites by in situ living free radical polymerization using a silicate-anchored initiator, J. Am. Chem. Soc. 121 (1999) 1615±1616. [62] J. Tudor, L. Willington, D. O'Hare, B. Royan, Intercalation of catalytically active metal complexes in phyllosilicates and their application as propene polymerization catalysts, Chem. Commun. (1996) 2031±2032. [63] M. Alexandre, P. Dubois, R. JeÂroÃme, M. Garcia-Marti, T. Sun, J.M. Garces, D.M. Millar, A. Kuperman, Polyolefin nanocomposites, WO Patent WO9947598A1 (1999). [64] M. Alexandre, P. Dubois, J.M. Garces, T. Sun, R. JeÂroÃme, in preparation. [65] P. Dubois, M. Alexandre, F. Hindryckx, R. JeÂroÃme, Homogeneous polyolefin-based composites, J. Macromol. Sci.: Rev. Macromol. Chem. Phys. C38 (1998) 511±565. [66] M. Alexandre, E. Martin, P. Dubois, M. Garcia-Marti, R. JeÂroÃme, submitted for publication.

61

62

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

[67] J. Heinemann, P. Reichert, R. Thomann, R. MuÈlhaupt, Polyolefin nanocomposites formed by melt compounding and transition metal catalyzed ethene homo- and copolymerization in the presence of layered silicates, Macromol. Rapid Commun. 20 (1999) 423±430. [68] Y.C. Ke, C.F. Long, Z.N. Qi, Crystallization, properties, and crystal and nanoscale morphology of PET±clay nanocomposites, J. Appl. Polym. Sci. 71 (1999) 1139±1146. [69] P.B. Messersmith, E.P. Giannelis, Synthesis and characterization of layered silicate-epoxy nanocomposites, Chem. Mater. 6 (1994) 1719±1725. [70] X. Kornmann, L.A. Berglund, J. Sterte, nanocomposite based on montmorillonite and unsaturated polyester, Polym. Eng. Sci. 38 (1998) 1351±1358. [71] T. Lan, T.J. Pinnavaia, Clay-reinforced epoxy nanocomposites, Chem. Mater. 6 (1994) 2216±2219. [72] T. Lan, P.D. Kaviratna, T.J. Pinnavaia, Mechanism of clay tactoid exfoliation in epoxy±clay nanocomposites, Chem. Mater. 7 (1995) 2144±2150. [73] C. Zilg, R. MuÈlhaupt, J. Finter, Morphology and toughness/stiffness balance of nanocomposites based upon anhydride-cured epoxy resins and layered silicates, Macromol. Chem. Phys. 200 (1999) 661±670. [74] Z. Wang, T.J. Pinnavaia, Hybrid organic±inorganic nanocomposites: exfoliation of magadiite nanolayers in an elastomeric epoxy polymer, Chem. Mater. 10 (1998) 1820±1826. [75] Z. Wang, T.J. Pinnavaia, Nanolayer reinforcement of elastomeric polyurethane, Chem. Mater. 10 (1998) 3769±3771. [76] R.A. Vaia, E.P. Giannelis, Lattice of polymer melt intercalation in organically-modified layered silicates, Macromolecules 30 (1997) 7990±7999. [77] A.C. Balazs, C. Singh, E. Zhulina, Modeling the interactions between polymers and clay surfaces through selfconsistent field theory, Macromolecules 31 (1998) 8370±8381. [78] A.C. Balazs, C. Singh, E. Zhulina, Y. Lyatskaya, Modeling the phase behavior of polymer/clay nanocomposites, Acc. Chem. Res. 8 (1999) 651±657. [79] Y. Lyatskaya, A.C. Balazs, Modeling the phase behavior of polymer±clay composites, Macromolecules 31 (1998) 6676±6680. [80] V.V. Ginzburg, A.C. Balazs, Calculating phase diagram of polymer±platelet mixtures using density functional theory: implication for polymer/clay composites, Macromolecules 32 (1999) 5681±5688. [81] R.A. Vaia, E.P. Giannelis, Polymer melt intercalation in organically-modified layered silicates: model predictions and experiment, Macromolecules 30 (1997) 8000±8009. [82] R.A. Vaia, H. Ishii, E.P. Giannelis, Synthesis and properties of two-dimensional nanostructures by direct intercalation of polymer melts in layered silicates, Chem. Mater. 5 (1993) 1694±1696. [83] R.A. Vaia, K.D. Jandt, E.J. Kramer, E.P. Giannelis, Kinetics of polymer melt intercalation, Macromolecules 28 (1995) 8080±8085. [84] R.A. Vaia, K.D. Jandt, E.J. Kramer, E.P. Giannelis, Microstructural evolution of melt intercalated polymerorganically modified layered silicates nanocomposites, Chem. Mater. 8 (1996) 2628±2635. [85] L.M. Liu, Z.N. Qi, X.G. Zhu, Studies on nylon-6 clay nanocomposites by melt-intercalation process, J. Appl. Polym. Sci. 71 (1999) 1133±1138. [86] M. Kato, A. Usuki, A. Okada, Synthesis of polypropylene oligomer-clay intercalation compounds, J. Appl. Polym. Sci 66 (1997) 1781±1785. [87] M. Kawasumi, N. Hasegawa, M. Kato, A. Usuki, A. Okada, Preparation and mechanical properties of polypropylene±clay hybrids, Macromolecules 30 (1997) 6333±6338. [88] N. Hasegawa, M. Kawasumi, M. Kato, A. Usuki, A. Okada, Preparation and mechanical properties of polypropylene±clay hybrids using a maleic anhydride-modified polypropylene oligomer, J. Appl. Polym. Sci 67 (1998) 87±92. [89] D. Wolf, A. Fuchs, U. Wagenknecht, B. Kretzschmar, D. Jehnichen, L. HaÈussler, Nanocomposites of polyolefin clay hybrids, in: Proceedings of the Eurofiller'99, Lyon-Villeurbanne, 6±9 September 1999. [90] M. Alexandre, G. Beyer, C. Henrist, R. Cloots, A. Rulmont, P. Dubois, in preparation. [91] M. Laus, O. Francesangeli, F. Sandrolini, New hybrid nanocomposites based on an organophilic clay and poly(styrene-b-butadiene) copolymers, J. Mater. Res. 12 (1997) 3134±3139. [92] S.D. Burnside, E.P. Giannelis, Synthesis and properties of new poly(dimethylsiloxane) nanocomposites, Chem. Mater. 7 (1995) 1597±1600. [93] S.J. Wang, C.F. Long, X.Y. Wang, Q. Li, Z.N. Qi, Synthesis and properties of silicone rubber organomontmorillonite hybrid nanocomposites, J. Appl. Polym. Sci. 69 (1998) 1557±1561. [94] A. Okada, K. Fukumori, A. Usuki, Y. Kojima, T. Kurauchi, O. Kamigaito, Rubber±clay hybrid Ð synthesis and properties, Polym. Prep. 32 (1991) 540±541. [95] A. Okada, A. Usuki, The chemistry of polymer±clay hybrids, Mater. Sci. Eng. C3 (1995) 109±115. [96] R.A. Vaia, S. Vasudevan, W. Krawiec, L.G. Scanlon, E.P. Giannelis, New polymer electrolyte nanocomposites: melt intercalation of poly(ethylene oxide) in mica-type silicates, Adv. Mater. 7 (1995) 154±156. [97] R.A. Vaia, B.B. Sauer, O.K. Tse, E.P. Giannelis, Relaxations of confined chains in polymer nanocomposites: glass transition properties of poly(ethylene oxide) intercalated in montmorillonite, J. Polym. Sci.: Part B Polym. Phys. 35 (1997) 59±67.

M. Alexandre, P. Dubois / Materials Science and Engineering 28 (2000) 1±63

[98] W. Chen, Q. Xu, R.Z. Yuan, Modification of poly(ethylene oxide) with polymethylmethacrylate in polymer-layered silicate nanocomposites, J. Mater. Sci. Lett. 18 (1999) 711±713. [99] H.R. Fischer, L.H. Gielgens, T.P.M. Koster, Nanocomposites from polymers and layered materials, Acta Polym. 50 (1999) 122±126. [100] K.A. Carrado, L.Q. Xu, In-situ synthesis of polymer±clay nanocomposites from silicate gels, Chem. Mater. 10 (1998) 1440±1445. [101] L. Mullins, N.R. Tobin, J. Appl. Polym. Sci. 9 (1965) 2993±3005. [102] Y. Yang, Z.-K. Zhu, J. Yin, X.-Y. Wang, Z.-E. Qi, Preparation and properties of hybrids of organo-soluble polyimide and montmorillonite with various chemical surface modifications methods, Polymer 40 (1999) 4407±4414. [103] A. Blumstein, Polymerization of adsorbed monolayers: II. Thermal degradation of the inserted polymers, J. Polym. Sci. A3 (1965) 2665±2673. [104] J. Lee, T. Takekoshi, E. Giannelis, Fire retardant polyetherimide nanocomposites, Mater. Res. Soc. Symp. Proc. 457 (1997) 513±518. [105] J.W. Gilman, Flammability and thermal stability studies of polymer layered-silicate (clay) nanocomposites, Appl. Clay Sci. 15 (1999) 31±49. [106] F. Dietsche, R. MuÈlhaupt, Thermal properties and flammability of acrylic nanocomposites based upon organophilic layered silicates, Polym. Bull. 43 (1999) 395±402. [107] J.W. Gilman, T. Kashiwagi, S. Lomakin, E.P. Giannelis, E. Manias, J.D. Lichtenhan, P. Jones, Nanocomposites: radiative gasification and vinyl polymer flammability, in: Proceedings of the 6th European Meeting on Fire Retardancy of Polymeric Materials (FRPM'97), University of Lille, France, 24±26 September 1997, pp. 203±221. [108] J.W. Gilman, T. Kashiwagi, J.E.T. Brown, S. Lomakin, Flammability studies of polymer layered silicate nanocomposites, SAMPE J. 43 (1998) 1053±1066. [109] F. Dabrowski, M. Le Bras, S. Bourbigot, J. Gilman, T. Kashiwagi, PA-6 montmorillonite nanocomposite in intumescent fire retarded EVA, in: Proceedings of the Eurofillers '99, Lyon-Villeurbanne, France, 6±9 September 1999. [110] C. Scherer, PA Film grade with improved barrier properties for flexible food packaging applications, in: Proceedings of the New plastics'99, London, 2±4 February 1999. [111] R.A. Vaia, G. Price, P.N. Ruth, H.T. Nguyen, J. Lichtenhan, Polymer/layered silicate nanocomposites as high performance ablative materials, Appl. Clay Sci. 15 (1999) 67±92. [112] J.W. Kim, S.G. Kim, H.J. Choi, M.S. Jhon, Synthesis and electrorheological properties of polyaniline±Na‡montmorillonite suspensions, Macromol. Rapid Commun. 20 (1999) 450±452. [113] M. Kawasumi, N. Hasegawa, A. Usuki, A. Okada, Liquid crystal/clay mineral composites, Appl. Clay Sci. 15 (1999) 93±108. [114] J.-H. Choy, S.-J. Kwon, S.-J. Hwang, Y.-I. Kim, W. Lee, Intercalation route to nano hybrids: inorganic/organic Ð high Tc cuprate hybrid materials, J. Mater. Chem. 9 (1999) 129±135. [115] V. Laget, C. Hornick, P. Rabu, M. Drillon, Hybrid organic±inorganic layered compound prepared by anion exchange reaction: correlation between structure and magnetic properties, J. Mater. Chem. 9 (1999) 169±174.

63