celulose Pevinil acetate nanocomposites

celulose Pevinil acetate nanocomposites

(Parte 1 de 2)

Cellulose Poly(Ethylene-co-vinyl Acetate) Nanocomposites Studied by Molecular Modeling and Mechanical Spectroscopy

Gregory Chauve, Laurent Heux, Rachid Arouini, and Karim Mazeau*

Centre de Recherches sur les Macromolecules Vegetales, CERMAV-CNRS, BP 53, 38041 Grenoble Cedex 9, France

Received February 16, 2005; Revised Manuscript Received April 28, 2005

Structure property relationships have been established at two different scales to examine reinforcing effects of nanocomposites made of cellulose whiskers and polyethylene-vinyl acetate (EVA) matrixes with different vinyl acetate contents. The role of the polymer structure on the work of adhesion as predicted by molecular modeling at the atomic scale and on the mechanical performance of nanocomposites observed by dynamic mechanical analysis at the macroscopic level is reported. Concordant results were obtained by the two approaches; both demonstrated a reinforcing effect that increases with the acetate content of the polymer. However, a leveling of this effect was observed at high acetate contents. A detailed picture of the interactions at the interface between the two species accessed by modeling gives a reasonable explanation of this unexpected phenomenon.


Cellulose fibers have long served as raw material in the textile and paper industriesor in compositematerialas filler.1 Recently, research focuses on applications with high added value, such as nanocomposites including cellulose nanocrystallites, also called whiskers.2,3 Among their interesting characteristics, cellulose whiskers are abundant and renewable. They exhibit also a low density as compared to mineral fillers (around 1.5 g cm-3), a high form factor of about 70 together with a high specific area of 150 m2 g. They have been shown to lead to remarkable reinforcing properties in such different matrixes as styrene-acrylate latex,4 starch,5 polyhydroxybutyrateoctanoate,6 or poly(ethyleneoxide).7 In most cases, the reinforcingeffects came from the percolating network of cellulose whiskers together with good interfacial compatibility between the matrix and the fillers. However, the relative role of these two contributions is not easy to assess.

In particular, the failure of our understanding of the interfacial phenomena comes from the absence of a precise description of the interactions between the filler and the matrix at the atomistic scale. So far, such a description relies on qualitative considerations, and the choice of the matrix is based mostly on trial and error. In contrast, the use of molecular models of the filler in contact with the polymer matrix can give new insights in terms of geometry and energies of the interaction.8-1 In this context, our recently developed models of the complex surface of cellulose were successfullyused to study the adsorption process of aromatic speciesof low molecularweight.12,13Such modelsare suitable to study the interface between cellulose and an amorphous polymer solid.

The present study reports complementary approaches that address the question of the rational choice of synthetic polymer matrixes for nanocomposites reinforced with cellulose whiskers. To this end, we use molecular modeling and dynamic mechanical analysis experiments to establish a reliable structure-property relationship that in turn allows for the design of an optimal polymer matrix that shows good adhesion properties on native cellulose. As vinyl acetate copolymers are used widely as wood and paper adhesives, their interactions with cellulose are of great industrial importance. In addition, the acetate content of ethylenevinyl acetate copolymers can be controlled easily during the synthesis giving the polymer matrixes that differ in their macroscopic polarity and surface energy. Such copolymers were then selected in this study as suitable for establishing a structure-property relationship. Cellulose whiskers were selected as fillers because the surface of such nanocrystals is well-defined at the atomic scale as shown by AFM.14-16 Such fillers allow realistic modeling of their surface properties, in contrast to common cellulose fibers.

Experimental Procedures

PolymerMatrixes.Differentpolymermatrixes(see Figure 1) were considered for the experimental and modeling studies: the homopolymers polyethylene (EVA0) and polyvinyl acetate (EVA100) and random copolymers of polyethylene-co-vinylacetatewith 12, 25, 40, and 75% w/w vinyl acetate. The acetate content and the random character of the distribution of the monomers within the copolymers were checked by 13C liquid-state NMR according to Sung and Noggle.25 Corresponding molar ratios and details of some physical properties are reported in Table 1.

Molecular Modeling. The calculations were carried out

using the all-atoms model using the Cerius2 molecular

10.1021/bm0501205 C: $30.25 © 2005 American Chemical Society Published on Web 06/21/2005

modelingpackage.18 The defaultparameterswere used except those explicitly mentioned. The pcff-300-1.0119 second generation force-field was employed. The bonded terms of the potential energy function exhibit quadratic terms to describe bond length and valence angles and also a threeterm Fourier transform for torsional angles. For the nonbonded contributions, the van der Waals interactions were calculated using a 9-12 Lennard-Jones function. Electrostatic interactions were accounted for through a classical Coulombpotential;the chargeequilibrationmethodwas used to calculate charges for each atom.20 The cut-in and cutoff distances were fixed at 10 and 1 Å, respectively. The minimum image convention was imposed in order not to duplicate nonbonded calculations.

The minimization uses the conjugate gradient procedure with the convergence criterion of the root-mean-square of the atomic derivatives of 0.05 kcal mol-1 Å-1. Molecular dynamic calculations were based on the canonical NVT ensemble (constant number of particles, volume, and temperature). The equations of motion were solved using the Verlet algorithm,21 with a time step of 1 fs. The system is coupled to a bath at T ) 400 K using Nose’s algorithm.2 Independentlyof the experimentaldata, this temperaturewas chosen because it allows a proper equilibrationof the system within a reasonable simulation time.

Cellulose. The initial two chain monoclinic unit cell of the Iâ allomorph23 was duplicated in all three dimensions, and a new triclinic supercell was redefined. This system was then equilibrated by a molecular dynamics run of 1 ns in the isothermal-isobaric NPT ensemble24 (constant number of particles, pressure, and temperature) and then optimized to reach its equilibrium geometry. As this superlattice possesses the (110) and (010) surfaces parallel to the faces of the periodic cube, it was redefined to expose the relevant (110) and (11h0) faces. The dimensions a and c of the final supercellare 28.3 and 21.20 Å, and the parameter,b, normal to the (11h0) surfaceis enlargedto providea sufficientvolume above the cellulose to insert the thin polymer film. The supercell includes five layers of cellulose chains, each composed of four glucose residues. All the chains are covalently connected to their respective periodic images to model infinite chain length. The inner layer, referenced as layer 3, was kept constrained during all the calculations to mimic the well-organized crystalline interior of the cellulose whisker. By contrast, the atoms of the cellulose chains belonging to the other four layers are unconstrained; this allowed the surface chains together with the first inner layer to adjust their geometry to account for the presence of the different polymer chains at the interface. The periodic boundary conditionsdefined an infinite surface of a cellulose whisker, exposing only its (11h0) surface.

Polymer. Each polymer single chain was generated accordingto the monomersequencesreportedin the literature from liquid-state NMR data.25 Unperturbed conformations were first generated following a Monte Carlo protocol coupled with an energy minimization procedure. The degree of polymerization(DP) was carefully chosen for each matrix (EVA0, EVA75, and EVA100), depending on the expected final film thickness (a 30-40 Å thickness has been shown to be reasonable according to preliminary computations) and on the experimental bulk density of the matrix (see Table 1). In fact, the height parameter (dimension b) of the final system should be large enough that a significant portion of the polymer is not perturbed by the interfaces with cellulose and, in addition,long-rangeinteractionsbetweenthe cellulose chains on both sides of the simulatedcube should be minimal if not null. Also, the heightparametershouldbe small enough for the whole system has to be computationally tractable. By contrast to the adjustable b parameter of the supercell, dimensionsa and c, which dependon the cellulosecrystalline organization,are strictly imposed. The selected chain lengths were DP ) 150, 295, and 385 for EVA100, EVA75, and EVA0, respectively, corresponding roughly to equivalent polymer volumes.

A thin polymer solid film was then created by inserting the generated polymer chain into a periodic cell having the same a and c parameters as the crystalline cellulose system and a b parameterlarge enough to includethe whole polymer chain in that direction and an impenetrable wall parallel to the a0c plane. Then, several compression cycles were applied; each consists of an initial energy minimization followed by molecular dynamics runs in the NVT ensemble. Then, the height parameter was diminished slightly before another compressioncycle was performed. Such cycles were repeated until the polymer density reached the desired experimental value of the bulk, reported in Table 1. A final 500 ps dynamic run (using the NVT ensemble) was performed to correctly equilibrate the polymer film.

Creationof the Interface.Havingcreatedand equilibrated both the amorphous polymer films and the crystalline cellulose, whose a and c dimensions are compatible, formation of the interface was straightforwardby inserting the film into the cellulose supercell above the cellulose chains. To avoid steric overlap, the distance between the polymer film and the cellulose chains was carefully selected as that of lowest energy. Finally, another 500 ps dynamic run was performed on the polymer/cellulose system; a view of a

Figure 1. Schematic representation of EVA polymers.

Table 1. Characteristics of the EVA Polymers Considered in the Present Study

XEa XVA polarity ç (293 K) Tm Tg density

EVA100 0 100 0.329 54 278** 1.191*X and X : molar ratio of ethylene and vinyl acetate monomers, respectively; ç: surface tension (mJ m ); T : melting temperature (K); T : glass transition temperature (K); density (g cm ). Data from Handbook. According to the supplier, Aldrich Chemical Co. According to our experiments.

2026 Biomacromolecules, Vol. 6, No. 4, 2005 Chauve et al.

typical system modeled is given in Figure 2. To check the reproducibility of the estimated data, two such systems were created for each of the three different matrixes, each possessing two independent interfaces, and results were then averaged over four values.

All calculations were performed at the Centre d’Experimentation et de Calcul Intensif, CECIC, Grenoble.

Aqueous Whiskers Suspension. Cellulose whisker suspensions in water were prepared by acid hydrolysis of the mantle of the tunicate, a sea animal, as described by Marchessault et al.26 The mantle of tunicate is formed of well-organized cellulosic microfibrils (tunicin) and therefore highly crystalline (essentially made up of the Iâ allomorph). The resulting rodlike particles (whiskers) exhibit a 10-20 nm width for a 1 ím average length, as estimated by transmission electron microscopy (TEM, CM200 Cryo, Philips), leading to an axial ratio of about 75.

Organic Whiskers Suspension. The aqueous suspension was then freeze-dried, and the resulting product was poured into toluene with vigorous stirring using an ultrahigh-speed blender (Ultra-Turrax from IKA) until a homogeneous suspension was obtained.

Film Processing. Polymer Matrixes. Polyethylene-covinylacetate(EVA)copolymersand a vinylacetate(EVA100) homopolymer were purchased from Aldrich Chemical Co. and used as received.

Four different polymer matrixes were considered for the experimentalstudy: polyethylene-co-vinyl acetate with 12% w/w vinyl acetate (EVA12), polyethylene- co-vinyl acetate with 25% w/w vinyl acetate (EVA25),polyethylene-co-vinyl acetate with 40% w/w vinyl acetate (EVA40), and polyvinyl acetate (EVA100). The details of some physical properties are reported in Table 1.


EVA25, EVA40, or EVA100) was dissolved in boiling toluene and mixed under strong magnetic stirring with the whiskers/toluenesuspension.The resultingmixed suspension was kept boiling for 30 min.

The suspension was then poured into a crystallizing dish, and the solvent was evaporated off overnight at room temperature. Residual toluene was removed in a vacuum oven at 303 K for 1 h. Small fragments were then placed in a mold and compressed with a Carver laboratory press at 13 800 kPa for 5 min at 423 K. The nanocomposites composed of whiskers and polymer contained 0-6 wt % tunicate whiskers, and their final thickness ranged between 0.5 and 1 m.

Dynamic Mechanical Analysis. Thermomechanical investigations were carried out with a Rheometrics RSA I spectrometer working in the tensile mode to measure the complex tensile modulus E* (i.e., the storage component E′ and the loss component E′′). Measurements were performed under isochronalconditionsat 1 Hz in a nitrogenatmosphere, and the temperature was varied between 150 and 500 K at a heating rate of3Km in-1 .

DifferentialScanningCalorimetry.Differentialscanning calorimetry (DSC) was performed using a DSC7 apparatus from Perkin-Elmer. Around 15 mg of sample was placed in a pressure-tightDSC pan. Each sample was heated from 150 to 400 K at a heating rate of 10 K min-1. The glass transition temperature (Tg) was taken as the inflection point of the specific heat increment at the glass-rubber transition, while the melting temperature (Tm) was taken as the peak temperature of the melting endotherm.

Results and Discussion

Molecular Modeling. Three polymer matrixes were considered in the modeling study: the two homopolymers EVA0 and EVA100 with only ethylene and vinyl acetate groups, respectively, and also EVA75 with an equal number of moles of each comonomer randomly distributed.

A typical modeled system is shown in Figure 2; it adequately represents a volume element of a nanocomposite in which the disorganized solid in the center of the figure correspondsto the matrix that is surrounded by five cellulose layers (only two of them are presented on both sides of the EVA polymer). Such a model gives access to geometrical and energetic characteristics of the adhesion process on the surface of the cellulose whisker.

The geometries of the polymers at the interface are analyzed first; we are particularly interested in the organization of the backbone of the polymer together with the orientation of their pendant acetate groups near the cellulose surface.

Density Profiles: Skeletal Organization of the EVA

Polymers. Figure 3 shows average density profiles of cellulose chains and of the backbone atoms of the polymer for the three equilibrated systems. The density has been calculated by considering layers of a 0.5 Å thickness along an axis normal to the surface of cellulose.

For the three systems, the crystalline organization of the cellulose chains does not appear significantly affected by the presence of the polymer. A limited overlap of the densities can be visualized for the most polar polymers. This interpenetration shows that the discontinuity between cel-

Figure 2. Typical molecular model of EVA75 (stick), viewed parallel to the cellulose longitudinal axis (balls). The original cell is duplicated in the direction parallel to the cellulose surface. Hydrogen atoms are omitted; the two vertical solid blue lines correspond to the original unit cell, and only two layers of cellulose on both parts of the matrix are represented.

Poly(Ethylene-co-vinyl Acetate) Nanocomposites Biomacromolecules, Vol. 6, No. 4, 2005 2027

lulose and polymer is fulfilled best at the atomic scale in the presence of vinyl acetate groups. A maximum density is also seen for the three polymers corresponding to the first layer in interaction with the cellulose surface. For EVA75 and EVA100,this maximumis rapidlydamped,and the inner part of the matrixes appears disorganizedin terms of density. The situation is slightly different for the EVA0 polymer, for which large amplitude fluctuations around the average density of about 1 gcm-3 are observed, which reveals a multilayered arrangement. The typical distance between two peaks roughly corresponds to the thickness of an ethylene monomer.

To better reveal the organization of the polymers, representative molecular models of the chain skeletal atoms are shown in Figure 4. As already seen in the density profiles, the inner part of EVA75 and EVA100 appears disorganized. Most probably this is related to the relative hindrance caused by the acetategroups.As a matter of fact, these two polymers are amorphous or poorly crystalline in their bulk state.

However, near the interface, some segments adsorb parallel to the cellulose surface; this also explains the first maximum observed in the density profiles for these two polymers.

The situation is slightly different for the ethylene homopolymer EVA0. A well-defined layer can be seen near the interface.This preorganizationis more or less transmitted to the inner part of the matrix, as seen in the density profile, probably related to the greater conformational freedom of the polyethylene chains as compared to polymers containing pendant acetate groups. This rough organization resembles the epitaxial crystallization that has been postulated for polypropylene near cellulose fiber.27 If our models are reliable, this crystallization should come then from geometrical constraints only.

Carbonyl Group Orientation. As the acetate containing polymersEVA75and EVA100is disorganizedin the absence of the cellulose surface, the partial layering observed in the density profiles near the interface could come from specific interactions between the polar groups of the matrix and the hydroxyl moieties of the cellulose. It can expected that electrostatic interactions, notably through hydrogen bonds, could play an important structural role at the interface, considering the numerous hydroxyl groups present at the surface of cellulose together with the polar nature of the acetate groups.

(Parte 1 de 2)