Relationship Between the Polymer/Silica Interaction and Properties of Silica Composite Materials AURE´LIEN GUYARD,1 JACQUES PERSELLO,2 JEAN-PHILIPPE BOISVERT,1 BERNARD CABANE3 1
CRPP, Universite´ du Que´bec a` Trois-Rivie`res, C.P.500,Trois-Rivie`res, Que´bec, G9A5H7, Canada
2
LCMI, Universite´ de Franche-Comte´, 16 route de Gray, 25030 Besanc¸on, France
3
PMMH, ESPCI, 10 rue Vauquelin, 75231 Paris, France
Received 29 August 2005; revised 5 January 2006; accepted 13 January 2006 DOI: 10.1002/polb.20768 Published online in Wiley InterScience (www.interscience.wiley.com).
Cast film composites have been prepared from aqueous polymer solutions containing nanometric silica particles. The polymers were polyvinyl alcohol (PVA), hydroxypropylmethylcellulose (HPMC) and a blend of PVA-HPMC polymers. In the aqueous dispersions, the polymer–silica interactions were studied through adsorption isotherms. These experiments indicated that HPMC has a high affinity for silica surfaces, and can adsorb at high coverage; conversely, low affinity and low coverage were found in the case of PVA. In the films, the organization of silica particles was investigated through transmission electron microscopy (TEM) and small-angle neutron scattering (SANS). Both methods showed that the silica particles were well-dispersed in the HPMC films and aggregated in the PVA films. The mechanical properties of the composite films were evaluated using tensile strength measurements. Both polymers were solid materials, with a high-elastic modulus (65 MPa for HPMC and 291 for PVA) and a low-maximum elongation at break (0.15 mm for HPMC and 4.12 mm for PVA). In HPMC films, the presence of silica particles led to an increase in the modulus and a decrease in the stress at break. In PVA films, the modulus decreased but the stress at break increased upon adding silica. Accordingly, the polymer/silica interC 2006 action can be used to tune the mechanical properties of such composite films. V
ABSTRACT:
Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 44: 1134–1146, 2006
Keywords: polyvinylalcohol; hydroxypropylmethylcellulose; neutron scattering; silica; structure; mechanical properties; nanocomposite
INTRODUCTION A growing interest is directed toward the development of high-performance composite materials in which nanosize inorganic particles are used to reinforce, instead of just fill, the organic matrix.
This work used the neutron beam of the Institut LaueLangevin Correspondence to: J.-P. Boisvert (E-mail: jean-philippe_
[email protected]) Journal of Polymer Science: Part B: Polymer Physics, Vol. 44, 1134–1146 (2006) C 2006 Wiley Periodicals, Inc. V
1134
The addition of solid particles deeply modifies and can significantly enhance the mechanical properties of the composite material, compared to the polymer material alone. Numerous studies have been published in recent years focusing on the use of carbon black and silica as reinforcing agents in elastomers.1–5 Much less have been published on glassy or semicrystalline polymer films. The reinforcing efficiency of fillers is a direct function of their volume fraction in the polymer matrix. For instance, the Guth and Gold approximation6 relates the elastic modulus E of the com-
POLYMER/SILICA INTERACTION AND PROPERTIES
posite to the volume fraction / of nondeformable filler particles: E=Eo ¼ 1 þ 2:5/ þ 14:1/2
ð1Þ
The first-order term is equivalent to the relative viscosity increase of a dilute suspension of spherical particles. The second-order term refers to the elastic interparticle interaction in the isotropically averaged stress field. This approximation remains valid, in the best case, for low to moderate deformations and typical loadings of filler up to / 0.35.7,8 It is then expected that high-volume fractions of particles will lead to composites with higher modulus, compared with the unfilled material. Equation 1 also implicitly assumes a homogeneous particle distribution within the organic matrix, maximizing direct contact with the filler surface.9 In such conditions, it has been shown that the composite modulus and strength can increase.10 On the other hand, heterogeneous filler distribution creates discontinuities where fractures are more likely to initiate.11 Other parameters are also important. The mechanical properties depend on the particle geometry, shape, size distribution, and porosity, as well as on the nature of the polymer/filler interaction.12–14 The porosity of fillers can allow interpenetration of the filler and the matrix and significantly reinforce composite materials.15–17 Finally, the number and nature of additional reticulation points created by the introduction of fillers have also been identified as important parameters regarding tensile properties of composites.4,15,18,19 However, it is not always clear in the literature whether the improvement is directly related to better polymer/filler interaction or whether to better distribution homogeneity of the filler due to better interaction. These interactions are often described by the so-called wettability of the filler by the matrix and are connected to the adhesion between the moieties. Poorly wetted filler surfaces lead to the creation of voids around the particles. These voids are weak points where failures initiate, as the material is under stress, then reducing its strength.20,21 To increase the filler wettability, coupling agents can be used. Over the years, many coupling agents have been commercialized for specific composites and applications, showing how important wetting is for the development of high-performance composites. The aim of this work was to determine how the mechanical properties of glassy composite materials can be related to their structure and to the Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
1135
polymer/surface interaction. For composite films that are made by solvent evaporation from mixed dispersions, the polymer surface interaction can influence the film properties directly, or indirectly through its effect on the organization of particles in the film. To decouple these effects, composite films were made with the same particles and three different polymers: polyvinylalcohol (PVA), hydroxypropylmethylcellulose (HPMC) and a mixture of HPMC and PVA where HPMC is preadsorbed on the particles. The polymer/surface interaction was investigated through measurements of adsorption isotherms in dilute dispersions. Composite films were then prepared through evaporation of the mixed dispersions, yielding filler volume fractions up to 30%. The structures of these composite films were studied with transmission electron microscopy (TEM) and small-angle neutron scattering (SANS). Finally, their mechanical properties were evaluated through uniaxial tensile measurements. The comparison of the properties of films with different polymer/surface interactions provides insights into the mechanisms by which composite films may become either softer or harder and brittle upon increasing the volume fraction of dispersed particles.
EXPERIMENTAL General All experiments were conducted in distilled and deionized water. The chemical reagents are all of analytical grade and were used without further purification. The solutions and suspensions were adjusted to pH 9 with NaOH.
Synthesis and Characterization of Silica The method of synthesis is fully described earlier.22 Basically, silica particles were grown from aqueous silicate solutions neutralized by nitric acid, as described by Iler.23 According to the transmission electron micrograph (not reported here, see Ref. 22), the particle hard sphere diameter is 27 nm. This value is consistent with the radius computed from the BET specific surface area (100 m2/g). The hydrodynamic diameter has been measured by dynamic light scattering (Malvern Zetasizer 4) and is 30 nm, and the polydispersity index is 1.07. The hydrodynamic size has also
1136
GUYARD ET AL.
¼ 0.494 reported elsewhere.24 The glass-transition temperature is around 75 8C. The (hydroxypropyl)methylcellulose (HPMC) was purchased from Fluka (cat. number 900465-3) with 64% of H, 29% of methoxy, and 7% of propylene oxide. The MW measured by viscosimetry is 1,275 kg/mol and RG ¼ 65 nm. The overlap concentration calculated from RG is estimated to be 1.9 g/L. The Flory parameter for HPMC is 0.46,25 and the PI is 6.12 according to GPC. The chemical structure of both polymers is presented in Figure 1. For convenience, their main properties are grouped together in Table 1. Adsorption Isotherms
Figure 1. Chemical structure of the polymer units for HPMC and PVA. For HPMC, R ¼ 64% H, 29% CH3, and 7% CH2 C(OH)H CH3.
been measured by viscosimetry through the Einstein relation: g=go ¼ 1 þ 2:5/
ð2Þ
where / is the volume fraction of silica particles. The latter two methods lead exactly to the same hydrodynamic size (30 nm). The suspension pH was set to 9 with NaOH, and the salt content was adjusted to 0.003 M with NaNO3.
Polymer Characterization The polyvinylalcohol (PVA) used in this work was purchased from Fluka (cat. number 9002-89-5). According to the supplier, this polymer has a molecular weight (MW) of 100 kg/mol and a degree of purity of 99% and hydrolysis of more than 99%. In the present study, the polymer was used without further purification. Viscosity measurements showed a MW of 107 kg/mol and a gyration radius RG ¼ 15 nm. The MW found by gel-permeation chromatography (GPC) is also 85.6 kg/mol, and the polydispersity index (PI) is 1.25. The overlap concentration for this polymer is 8 g/L in water. The Flory parameter (v) in semidilute condition was measured by osmometry. The experimental value was 0.499, which agrees well with v
SiO2 dispersions at initial pH 9 were mixed with increasing concentrations of polymer solutions at same pHs. Five days were allowed for equilibrium at the end of which the suspensions were centrifuged and the total organic carbon (TOC) of the supernatant was measured with a Carbon analyzer instrument (Dohrmann). Only experimental data with (initial TOC-equilibrium TOC)/initial TOC > 0.2 were taken as significant. Calibration curve was established with known concentrations of polymer. The standard deviation for three consecutive measurements was lower than 5%. Film Formation Films about 80–100-lm thick were prepared by mixing SiO2 suspensions (/ ¼ 0.02, pH 9) with the corresponding polymer solution (5% w/w, pH 9) in appropriate proportion to end up with solid mean volume fractions (SiO2/(SiO2þpolymer)) ranging from /mean ¼ 0.03 to 0.3 once the solvent has been evaporated. The evaporation was achieved at room temperature and free atmosphere and completed within 8–10 h. Before evaporation, the polymer surface coverage of the SiO2 is well above saturation. Composite films were also prepared with a blend of both polymers. They were prepared in two steps. First, SiO2 suspensions (pH 9) were Table 1. Main Characteristics of the Polymers Used in this Study
HPMC PVA
RG (nm)
c* (g/L)
v
MW (kg/mol)
PI
65 15
1.9 8.0
0.46 0.494
1275 107
6.12 1.25
Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
POLYMER/SILICA INTERACTION AND PROPERTIES
mixed with HPMC polymer solution (pH 9) as described earlier. The HPMC concentration was adjusted in such a way that the surface coverage reached saturation but still keeping the equilibrium concentration close to zero. These two conditions are found on the isotherm (see later) where the data are simultaneously high on the y-axis and close to zero on the x-axis. A few days were allowed to ensure complete adsorption. Second, the PVA polymer solution (pH 9) was added to the HPMC-SiO2 suspension to finish with the same /mean as mentioned earlier, once evaporation was completed. Electron Microscopy Transmission electron microscopy (TEM) sample preparation consisted of immobilizing the films in a resin and slicing the film with an ultramicrotome. The cross section was fixed onto a coated carbon grid. TEM examination was performed using a Philips instrument operating at 120 kV. Neutron Scattering The neutron scattering experiments were performed at the ILL institute (Grenoble, France) on the D11 instrument. The neutron wavelength ˚ (610%). Detector distance was 5 and was 6 A 20 m providing an experimental range of 0.002 ˚ 1 < Q < 0.075 A ˚ 1, where Q is the modulus of A the scattering wavevector. The raw data were normalized and corrected for background with an unfilled polymer film. Tensile Strength Measurements Stress–strain measurements of dumbbell-shaped samples in uniaxial extension were carried out on an Instron testing machine at extensions lower than 6%. The nondeformed thickness of each sample was measured with a micrometer. At least five measurements per sample were performed with a crosshead speed of 5 cm/min in a climate room (23 8C and 50% humidity). Slipping of samples from the clamps was avoided using adhesive on the clamps. From values of nominal stress (r) and elongation (k), the modulus (E) was calculated with26: r¼
E 1 ½k 2 3 k
1137
culated with eq 3 by the linear regression of at least 10 data points. The stress at break (rb) and strain at break (eb ¼ k 1) were measured, and average and standard deviation were calculated. The elongation at break of the neat polymer films is 0.15 for HPMC and 4.12 for PVA.
RESULTS Adsorption Isotherms In aqueous dispersions, the interaction of polymers with particle surfaces can be determined through measurements of adsorption isotherms. The adsorption isotherms of HPMC and PVA on the nanometric silica particles are presented in Figure 2. They are strikingly different. The adsorption isotherm of HPMC has a very sharp rise at lowpolymer concentrations, indicating that the macromolecules have a high affinity for the silica surfaces at pH 9. The plateau coverage is 0.3 mg/m2, which corresponds to nearly full coverage. The adsorption isotherm of PVA shows that the macromolecules have a very low affinity for the silica surfaces at pH 9. The plateau coverage is below 0.01 mg/m2, which corresponds to 0.16 macromolecule, or 400 monomers per silica particle. This is an extremely low coverage; as a comparison, poly(ethylene oxide), which has monomers of the same size, would bind 20,000 monomers to the same surface area.27 In film-forming solutions, before evaporation, the concentration of polymers was at least two decades above the condition for surface saturation. Moreover, the polymer concentration was also at least four times above the overlap concen-
ð3Þ
where k ¼ L/L0, L0, and L being the initial and final length of the sample, respectively. E was calJournal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
Figure 2. Adsorption isotherms of HPMC and PVA on silica particles at pH 9.
1138
GUYARD ET AL.
Observations of Film Structure
Figure 3. TEM micrographs of (A) silica particles embedded in HPMC film matrix, (B) silica particles embedded in PVA film matrix, (C) HPMC-coated silica embedded in PVA film matrix. The bar is 500 nm and /mean is 0.054. Films are prepared at pH 9.
TEM was used to investigate the effect of polymer type and volume fraction of SiO2 (on a dry basis) on the structure of the composite. For convenience, only the micrographs of the films at volume fraction /mean ¼ 0.054 are reported in Figures 3(A–C). According to Figure 3, important differences of composite structures are observed with the different polymer matrices. With the HPMC matrix, the silica particles are mostly dispersed even though some doublets are present, while the primary particles have clearly collapsed into large, spherical, and dense clusters with the PVA and HPMC-PVA matrices. However, great care must be taken when interpreting two-dimension images resulting from the projection of a three-dimension object. This is especially true when high-silica contents are involved and when the objects observed are smaller than the thickness of the ultramicrotome slice. To support TEM observations, the composite structure at increasing mean volume fractions (/mean) has been investigated with small-angle neutron scattering (SANS). This technique gives access to the mean structural properties of the material in the three dimensions and is a good complement to electron microscopy. The SANS data for the silica-HPMC composites made with three volume fractions of silica are presented in Figure 4. The three spectra have been shifted vertically, so that they have the same intensities at high Q, where the scattering results from intraparticle interferences only. With this adjustment, the intensity scattered at low Q by the dispersions with higher volume fractions is depressed with respect to that of the dispersion with the lowest volume fraction. This depression results from interparticle interferences; it is a classical feature of dispersions with repulsive interactions. Consequently, the spectra were fitted with the calculated scattering curves for repelling hard spheres. For monodispersed spherical particles, it is well-known that the scattered intensity I(Q) can be expressed as the product of a single particle form factor, P(Q), and an interparticle structure factor, S(Q): IðQÞ PðQÞSðQÞ
tration, that is the polymers crossed from the semidiluted to the concentrated regime during the process of solvent evaporation.
ð4Þ
The structure factor S(Q) is determined by the pair correlation function of the particles, g(r), and Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
POLYMER/SILICA INTERACTION AND PROPERTIES
Figure 4. SANS data for SiO2-HPMC films at /mean ¼ 0.054 (*), /mean ¼ 0.095 (~) and / mean ¼ 0.287.(). The lines are the theoretical spectra for a liquid of repelling hard spheres at the same volume fraction.
1139
replacing the form factor P(Q) with an average form factor, Pav(Q), calculated by summing the form factors of all the particles in the distribution. An analytical expression of Pav(Q) exists in the case of the Schultz distribution (Appendix). The intensity calculated in this way provides a good fit to the spectra of the dispersions with silica volume fractions / 0.095 (Fig. 4). However, for / ¼ 0.287, the model cannot fit the experimental data at small angles: the actual depression of the intensity at small Q is much less than predicted. This is a clear indication that the interparticle correlations are not those of monodisperse hard spheres. There are two possible explanations for this: either the particles do not repel (they are partly aggregated) or their size distribution is such that small particles can be located within the spaces that separate larger ones. The silica-PVA films have also been investigated by neutron scattering. The spectra are presented in Figure 5. They all show a – 4 slope at low Q values, followed by a strong correlation peak positioned at a Q value corresponding to the particle diameter. As described elsewhere,22,29,30
by the number of particles per unit volume, n : Z SðQÞ ¼ 1 þ 4pn
1
½gðrÞ 1r2
0
sinðQrÞ dr Qr
ð5Þ
Liquid-state theory provides an analytical expression for S(Q), in the case of hard spheres, through the Wertheim solution of the Percus-Yevick closure approximation for the Ornstein-Zernike equation.28 In this calculation, the variables are the volume fraction / and the hard-sphere particle radius R. The latter has been estimated from TEM micrographs and BET to be R ¼ 13.5 nm. Moreover, the single particle form factor P(Q) of monodisperse spheres is:
3j1 ðQRÞ 2 PðQÞ QR
ð6Þ
where ji(QR) is the first-order Bessel function: J1 ðQRÞ ¼ ðsinðQRÞ QR cosðQRÞÞ=ðQRÞ2
ð7Þ
The intensity calculated in this way does not fit the experimental spectra, because the silica particles are actually not monodisperse in size. For polydisperse hard spheres, eq 4 is no longer valid. However, an approximate expression of the scattered intensity can still be obtained by Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
Figure 5. SANS data for SiO2-PVA films at / mean ¼ 0.054 (*), / mean ¼ 0.095 (l) and / mean ¼ 0.287 (*). The straight line indicates a –4 slope in log–log scales.
1140
GUYARD ET AL.
where N is the number of particles per cluster, S(Q) is the structure factor related to the local volume fraction (/local) of the primary particles inside the clusters, Pc(Q) is the form factor for clusters with a radius Rc. In eq 8, /local is taken as an adjustable parameter in the analytical solution of the Percus-Yevick model to fit the experimental data. The other parameters needed to solve eq 8 are estimated by TEM (the radius of clusters (Rc) and the average radius of the primary particles (R ¼ 13.5 nm)). An example of such modeling is presented in Figure 7 for the SiO2-HPMC-PVA composite. The best fitted /local values for the SiO2-PVA and SiO2-HPMC-PVA are reported in Table 2. The data show that the cluster structure in the PVA matrix can be modified by the presence of preadsorbed HPMC. It appears that HPMC makes the clusters more open (lower /local) for a given /mean compared to the situation with PVA only. The cluster size can also be computed from the SANS curves shown in Figures 5 and 6 by the
Figure 6. SANS data for SiO2-HPMC-PVA films at /mean ¼ 0.054 (*), /mean ¼ 0.095 (l) and /mean ¼ 0.287 (þ). The straight line indicates a –4 slope in log–log scales.
these features indicate that dense clusters are present within the polymer matrix. It has been concluded in previous work31 that the primary particles aggregate to form dense clusters. Finally, the influence of preadsorption of HPMC prior to mixing with the PVA on the final composite structure has been examined. According to the results presented in Figure 6, the presence of preadsorbed HPMC on the silica surface is not sufficient to prevent the microphase separation to occur at some stage after mixing with PVA. The SANS data are similar to those observed in the silica-PVA composite and typical of liquid-like finite-size clusters. A quantitative analysis of the results presented in Figures 5 and 6 can be derived by considering large clusters of silica particles dispersed within the polymer matrix. These clusters have two length scales: the first one for the primary particles inside the clusters and the second one for the overall size of the clusters. The model can then be described by eq 831,32: IðQÞ PðQÞ½ðN 1ÞPc ðQÞ þ SðQÞ
ð8Þ
Figure 7. Fit of SANS data (symbols) for the SiO2HPMC-PVA film (/mean ¼ 0.095) using eq 8 (full line). The local volume fraction in the clusters (/l) is taken as an adjustable parameter. The concordance between the correlation peaks (fit vs. experimental) is possible only when /l is set to 0.50. The data reported in Table 2 were found by the same procedure applied to corresponding films. Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
POLYMER/SILICA INTERACTION AND PROPERTIES
Table 2. Local Volume Fraction in Clusters (/local) for Different Mean Volume Fractions of Silica Particles in the Composite /mean
0.054
0.095
0.287
SiO2-HPMC SiO2-PVA SiO2-PVA-HPMC
No clusters 0.50 0.35
No clusters 0.60 0.50
ND 0.74 0.60
method described previously.31 The results are presented in Table 2. At a given /mean, the clusters are larger when formed in the HPMC-PVA system, compared with PVA only. Tensile Properties When perfectly dispersed and wetted in the matrix, filler particles can maximize the possible attachment points with the polymer, that is the apparent crosslink density of the matrix, and then maximize the hardness and modulus.10,18,20,33 The surface interaction with the polymer matrix is critical with these fillers since their high-surface area can lead to a very large number of attachment points with the matrix, and the tensile properties are expected to increase with the filler volume fraction. Such an improvement has been reported elsewhere.15,17,18,34 On the other hand, studies on silica fillers have also shown a monotonous reduction of strain and stress at break.20 In the latter case, a bad wetting of the surface by the polymer is expected to create voids in the vicinity of the particles and acts as weak points. A typical stress–strain curve is reproduced for each composite in Figure 8. The failure mode appears to be ductile for PVA and PVA-HPMC composites, while it is fragile for HPMC composite (insert in Fig. 8). This interpretation is supported by SEM images of the stretched ends not shown here (see Ref. 22 for PVA composites). Curves at higher /mean keep the same failure mode but with a shorter elongation at break, except at very high /mean where the failure mode becomes fragile. The modulus and stress at break were determined from these stress–strain curves. It is noteworthy that tensile tests can be very sensitive to randomly distributed surface defects of the film. Accordingly, the measurements where repeated at least five times (most of the time, up to 10 times) for each sample. According to the standard deviations reported on the following Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
1141
figures, the background noise on tensile tests coming from such random defects is not too important. Having this in mind, the influence of the silica volume fraction on the tensile modulus of the three composites has been calculated from eq 3 As shown in Figure 9, the modulus follows quite nicely the Guth and Gold relation (eq 1) at low to intermediate /mean. However as /mean reach a critical value, the modulus falls down to values that can be even lower than the unfilled polymer. Apparently, there is no direct relationship between the behavior of the stress at break (rb) against /mean and the surface interaction with the polymers. As reported in Figure 10, rb can increase (PVA), remain constant (HPMC-PVA) or decrease (HPMC) with /mean. The relationships between the surface properties and structural and tensile properties are debated below.
DISCUSSION Polymer/Surface Interaction and Composite Structure According to SANS and TEM, a phase separation occurs upon evaporation in the case of aqueous SiO2-PVA dispersions, while the particles remain well dispersed in the HPMC solutions [Figs. 3(A,B)].
Figure 8. Typical stress–strain curves of the filled composites (/mean ¼ 0.095) used for the determination of the failure stresses and elongations. SiO2-HPMC (D), SiO2-PVA (^), and SiO2-HPMC-PVA (). The stress at break for the neat polymers are ro ¼ 69 MPa (PVA), 75 MPa (HPMC), and 73 MPa (PVAþHPMC).
1142
GUYARD ET AL.
Figure 9. Elastic modulus of the filled composite relative to the unfilled polymer for SiO2-HPMC (D), SiO2-PVA (^), and SiO2-HPMC-PVA (l). The modulus is computed from eq 3. The solid line is computed through the Guth-Gold model (see eq 1 in text). Dotted lines are only guide to eyes. Eo ¼ 65 MPa (HPMC), 291 MPa (PVA), and 171 MPa (PVAþ HPMC).
As described in textbooks on the subject earlier,35–37 the state of dispersion depends on the balance of repulsive and attractive forces acting on the particles. In aqueous dispersions containing ionized particles, such as silica, and neutral macromolecules, at least four interaction forces must be considered; the long-range electrostatic force (repulsive), the short-range van der Waals forces (attractive), the depletion force (attractive), and the steric force (repulsive). In the present experimental conditions, the electrostatic force is the same for all systems at a given evaporation stage. The same assumption is made for the van der Waals forces. Accordingly, only different depletion attraction and/or steric protection forces can explain the different behaviors of the systems investigated here. The relative magnitudes of these two forces depend on the polymer/surface interactions, which have been measured through the adsorption isotherms (Fig. 2). In the aqueous SiO2-PVA system, the coverage of silica surfaces is extremely low. Accordingly, most PVA macromolecules are repelled by the silica surfaces, and a water layer depleted of macromolecules surrounds each particle. In such conditions, the depletion forces become strong when the polymer concentration rises due to evaporation,38 and they promote the aggregation of the particles.31 To the contrary, in the aqueous SiO2-HPMC system, the macromolecules are adsorbed at high
coverage by the silica surfaces. Accordingly, a polymer layer made of the loops and tails of the adsorbed macromolecules surrounds each particle.39 In such conditions, repulsive osmotic pressures due to these layers prevent the aggregation of the macromolecules. Quantitatively, the protection (steric force) is expected to increase with the square of surface coverage,37 which is much higher with HPMC than with PVA (Fig. 2). As HPMC is preadsorbed on the surface, both effects combine; PVA still depletes the particles, but HPMC prevents the particles to collapse as severely as observed in the SiO2-PVA systems. According to the data reported in Table 2 and to TEM, HPMC in the SiO2-HPMC-PVA systems keeps the particles farther apart, although it cannot prevent the formation of clusters. The higher surface affinity of HPMC for the silica prevents it to be displaced from the surface by PVA, in spite of the high-bulk concentration of PVA. Such a behavior is expected since high MW chains preferentially adsorb and are not likely to be displaced by lower MW chains later on.39 Polymer/Surface Interaction and Mechanical Behavior at Small Deformations The mechanical properties of the polymer films are substantially modified by the addition of silica particles; the effects are not the same at small and large deformations. At small deformations, the SiO2-HPMC composites have a higher elastic modulus than the pure
Figure 10. Stresses at break of the filled composite relative to the unfilled polymer for SiO2-HPMC (D), SiO2-PVA (^), and SiO2-HPMC-PVA (). The lines are only guide to eyes. ro ¼ 69 MPa (PVA), 75 MPa (HPMC), and 73 MPa (PVAþHPMC). Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
POLYMER/SILICA INTERACTION AND PROPERTIES
polymer film. For composites with a silica volume fraction /mean < 0.20, the variation of relative modulus with /mean follows the Guth and Gold relation quite nicely (Fig. 8). Accordingly, nondeformable particles that are well dispersed and have strong interactions with the polymer do reinforce the glassy polymer matrix, in a way that matches the theoretical predictions. For composites with higher silica volume fractions (/mean > 0.20), the reinforcement is less than predicted, presumably because the particles are no longer well dispersed. Indeed, the SANS spectrum of the composite made at /mean ¼ 0.287 has a higher intensity at low Q than the theoretical spectrum for repelling particles (Fig. 4). The SiO2-HPMC-PVA composites show the same reinforcement of the modulus, despite the fact that the particles are partly aggregated. According to the analysis of SANS spectra (Table 2), the silica clusters in the composites made at /mean < 0.20 still contain some polymer (/local 0.50), and therefore all particles in a cluster do interact with the matrix. Consequently, the reinforcement of the elastic modulus depends mainly on proper mechanical interactions between the particles and the matrix; aggregated particles do provide the same reinforcement, provided that the aggregates are not dense. Conversely, the smaller reinforcement obtained at the highest silica volume fraction (/mean ¼ 0.287) results from the fact that the aggregates are nearly dense (/local ¼ 0.60), and therefore not all particles interact with the matrix. The SiO2-PVA composites only show a modest reinforcement up to /mean ¼ 0.09. At higher volume fractions, they are softer than the pure polymer matrix. Therefore, the addition of silica creates some weak regions in the PVA matrix. There are three possible explanations for this softening as the particle content increases. These possibilities are either related to modification of the degree of crystallinity of the polymer, or related to the formation of weak regions in the vicinity of the silica/PVA interface. PVA is a semicrystalline polymer at room temperature, with small crystallites immersed in a glassy matrix. The crystallites cause a crosslinking of the polymer, and contribute to its mechanical strength. Any modification of this crystallinity is therefore likely to change the tensile properties. As reported elsewhere, the crystallinity of PVA can be significantly deteriorated by the presence of fillers, such as Montmorillonite,34 and also silica that has been precipitated in situ Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
1143
through a sol–gel process.40–42 The gradual disappearance of polymer crystallites as the filler content increases could lead to degradation of the mechanical properties of the polymer matrix. To examine this possibility, X-ray diffraction has been conducted on the PVA-SiO2 composites. According to results not shown here for the sake of simplicity, no significant difference in crystallinity was found between the neat polymer and the composites, regardless of silica content. Consequently, the reduction of the modulus at highsilica volume fraction (/mean ˜ 0.09) cannot be explained by arguments related to the crystallinity of PVA. The second explanation would involve the presence of weak regions in the vicinity of the silica/PVA interfaces. Indeed, during the formation of PVA films, each silica particle maintains near its surface an aqueous layer that is depleted of PVA macromolecules. At high-water contents, this is demonstrated by the adsorption isotherm, which indicates a very low coverage of the surfaces by the macromolecules. At lowwater content, this depletion is the driving force for the aggregation of the particles. At the end of the film-formation process, the films are in equilibrium with ambient humidity, and therefore each silica surface must retain a layer of adsorbed water molecules. During strain, the weak PVA/surface interaction and coverage would make it easy for the macromolecules to desorb from the surface. Consequently, the silica/ polymer interfaces are weak regions of the material, which can cause the elastic modulus to fall below that of the neat polymer. A third possible explanation would involve failure of the silica/silica interfaces within a cluster. Indeed, the depletion of PVA macromolecules from the vicinity of silica surfaces may cause the particles within a cluster to be either in direct contact, or separated by a water layer only. These particle–particle contacts may be weaker than the glassy polymer matrix. As the silica volume fraction is increased above /mean ¼ 0.09, the local silica volume fraction within a cluster reaches /local ¼ 0.60, and the clusters may be full of such weak interparticle contacts. In this case, even a small deformation of the matrix would be enough to split a cluster apart. Some evidence of such cluster failure has been presented previously.22 If the clusters are no longer nondeformable fillers, then the Guth and Gold relation does not apply, and the modulus of the composite can fall below that of the neat polymer.
1144
GUYARD ET AL.
Polymer/Surface Interaction and Mechanical Behavior at Large Deformations At large deformations, the effects of the fillers are in opposite directions to those observed at small deformations. An inverse dependence between rb and / is usually taken as an indication of a poor surface wetting of the filler particles by the matrix. In such a situation, a decohesive failure is likely to occur at the interface. This behavior could have been expected in PVA composites, due to the low surface coverage of silica by PVA. However, just the reverse has been observed: the stress at break increases with silica content in the case of PVA, and it decreases in the case of HPMC, which has a high affinity for silica. Hence, other explanations must be sought. The failure properties in a composite material are the result of a balance between two effects: the heterogeneous stress distribution within the sample and the disorder that decreases the crack propagation within the sample. The first contribution seems to be dominant for samples having good particle dispersion state (i.e., there is high polymer/silica affinity). The second contribution could explain the higher rb values measured for poorly dispersed samples. (i.e., when there is low affinity between the polymer chains and the silica surface).43 In the case of the SiO2-HPMC composites, the magnitude of the stress at break is below that of the neat polymer, and it falls regularly when the silica volume fraction rises. This is caused by a decrease in the maximum extension at break, indicating that cracks propagate more easily. According to the Griffith criteria for crack propagation, there are two possible explanations for this: either the material has become less ductile or the size of defects at the surface of the film has increased. The SiO2-PVA composites show a very different behavior: at small silica volume fractions, the stress at break rises slowly with silica content; and then, at the highest volume fraction, it drops well below that of the pure polymer. The initial rise indicates that the material has become more ductile, making crack propagation more difficult. This is in line with the decrease in the elastic modulus at intermediate volume fractions. An alternative explanation would be similar to the mechanism proposed by Witten in the case of elastomers: as the matrix in the loaded area is stretched thin, the clusters are pushed closer together in the cross-direction and reach the point
where the silica particles get in contact with each other. From then on, the material resists to further stretching, which keeps rb high. The sudden drop at the highest volume fractions is presumably caused by the presence of very large defects in the material. Indeed, SANS measurements indicate that the average cluster sizes have become much larger (670 nm) at this volume fraction. Moreover, transmission electron micrographs22 show that in this condition the clusters span all over the composite volume and create an infinite three-dimensional network. In this extreme situation, the material may not be silicaembedded into a PVA matrix anymore, but rather PVA-embedded in a silica network, which explains the strong reduction of rb. Finally, the SiO2-PVA-HPMC composites show an intermediate behavior: at small silica volume fractions, the stress at break remains the same as that of the polymer mixture; this may be due to a compensation of the effects operating in the pure polymers. At large silica volume fractions, there is a catastrophic drop in the stress at break, which is clearly caused by the presence of very large defects. Indeed, SANS measurements indicate that the average cluster sizes have become quite large (1000 nm) at this volume fraction.
CONCLUSIONS Polymer/particle interactions have a direct effect on the structures of glassy or semicrystalline composites: favorable interactions prevent particle aggregation and lead to a good dispersion of the filler particles inside the polymer matrix, whereas poor interactions cause the filler particles to aggregate during the evaporation of the polymer-filler-solvent mixture. Polymer/particle interactions also have a direct effect on the mechanical behavior of composites at small deformations: favorable interactions increase the elastic modulus above that of the pure polymer, because well-dispersed, nondeformable particles reinforce the matrix; conversely, poor interactions produce weak layers near the polymer/particle interfaces, which cause the elastic modulus to drop below that of the pure polymer. On the other hand, there is no direct relation between polymer/particle interactions and the mechanical behavior of the composites at large deformations. Composites with poor interactions can have a stress at break that exceeds that of Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
POLYMER/SILICA INTERACTION AND PROPERTIES
the matrix, and conversely. The relations are probably indirect, as the polymer/particle interactions determine the structures of the composites, and the way in which these structures change upon deformation. The authors would like to acknowledge Peter Lindner and Joanes Zipfel (Institut Laue-Langevin, Grenoble) for technical support on the D11 instrument and Agne`s Lejeune from UQTR for supplying the TEM micrographs. NanoQue´bec is acknowledged for financial support. Finally, J.-P. Boisvert expresses its gratitude to Reviewer #2 for constructive comments.
APPENDIX The analytical equation for the weighted average P(Q)S is given by44,45: 6
PðQÞs R ðZ þ 1Þ6 aZþ7 GðQÞ
ðA1Þ
where a¼
Zþ1 ; QR
GðQÞ ¼ aðZþ1Þ ð4 þ a2 ÞðZþ1=2Þ cos½ðZ þ 1Þ arctan ð2=aÞ þ ðZ þ 2ÞðZ þ 1Þ faðZþ3Þ þ ð4 þ a2 ÞðZþ3Þ=2 cos ½ðZ þ 3Þ arctan ð2=aÞg 2ðz þ 1Þ ð4 þ a2 ÞðZþ2Þ=2 sin½ðZ þ 2Þ arctan ð2=aÞ
REFERENCES AND NOTES 1. Donnet, J. B.; Voet A., Carbon Black Physics, Chemistry and Elastomer Reinforcement; Marcel Dekker, New York, 1976. 2. Leblanc, J. L. Prog Polym Sci 2002, 27, 627–687. 3. Boonstra, B. B.; Cochrane, H.; Dannenberg, E. M. Rubber Chem Tech 1976, 48, 558–576. 4. Voet, A.; Morawski, J. C.; Donnet, J. B.; Rubber Chem Tech 1977, 50, 342–355. 5. Kelly, A. A Concise Encyclopedia of Composite Materials, New York, 1994. 6. Guth, E.; Gold, O. Phys Rev 1938, 53, 322–328. 7. Christensen, R. M. Mechanics of Composite Materials; Wiley: New York, 1979. 8. Chen, H.-S.; Acrivos, A. Int J Solids Struct 1978, 14, 331. 9. Wu, C. L.; Zhang, M. Q.; Rong, M. Z.; Friedrich, K. Compos Sci Technol 2002, 62, 1327–1340. Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
1145
10. Westerman, S.; Kreitschmann, M.; PyckhoutHintzen, W.; Richter, D. Macromolecules 1999, 32, 5793–5802. 11. Bokobza, L.; Garbaud, G.; Mark, J. E. Chem Mater 2002, 14, 162–167. 12. Berthelot, J. M. Mate´riaux composites, Comportement me´canique et analyse des structures, 3e e´dition,Editions Tec & Doc 1999. 13. Berriot, J.; Montes, H.; Martin, F.; Mauger, M.; Pyckhout-Hintzen, W.; Meier, G.; Frielinghaus, H. Polymer 2003, 44, 4909–4919. 14. Nakamura, Y.; Okabe, S.; Yoshimoto, N.; Iida, T. Polym Polym Compos 1998, 6, 407–414. 15. Jianfen, J.; Salovey, R. J Mater Sci 1999, 34, 4719–4726. 16. McCarthy, D. W.; Mark, J. E.; Schaeffer, D. W. J Polym Sci 1998, 36, 1167–1189. 17. Deng, Q.; Hahn, J. R.; Stasser, J.; Preston, J. D.; Burns, G. T. Rubber Chem Technol 2000, 73, 647– 665. 18. Osman, M. A.; Atallah, A.; Mu¨ller, M.; Suter, U. W. Polymer 2000, 42, 6545–6556. 19. Donnet, J. B. Compos Sci Technol 2003, 63, 1085– 1088. 20. Nunes, R. C. R.; Fonseca, J. L. C.; Pereira, M. R. Polym Test 2000, 19, 93–103. 21. Nakamura, Y.; Okabe, S.; Iida, T.; Nagata, K.; Nigo, H. Polym Polym Compos 1999, 7, 33–40. 22. Boisvert, J.-P.; Persello, J.; Guyard, A. J Polym Sci 2003, 41, 3127–3138. 23. Iler, R. K. The Chemistry of Silica;Wiley, New York, 1979. 24. Lewandowska, K.; Staszewska, D. U.; Bohdanecky, M. Eur Polym J 2001, 37, 25–32. 25. Polymer Data Handbook, Oxford University Press, 1999. 26. Ferry, J. D. Viscosielastic Properties of Polymers; Wiley: New York, 1970. 27. Lafuma, F.; Wong, K.; Cabane, B. J Colloid Interface Sci 1991, 143, 9–21. 28. Ornstein ??? 29. Pontoni, D.; Narayanan, T.; Petit, J.-M.; Gru¨bel, G.; Beysens, D. Phys Rev Let 2003, 80, 188301-1– 4. 30. Pontoni, D.; Narayanan, T. J Appl Cryst 2003, 36, 787–790. 31. Persello, J.; Boisvert, J.-P.; Guyard, A.; Cabane, B. J Phys Chem-B 2004, 108, 9678. 32. Wong, K.; Lixon, P.; Lafuma, F.; Lindner, P.; Aguerre Charriol, O.; Cabane, B. J Colloid Interface Sci 1992, 153, 55–68. 33. Bokobza, L. Macromol Symp 2001, 171, 163–170. 34. Strawhecker, K. E.; Manias, E. Chem Mat 2000, 13, 2943. 35. Russel, W. B.; Saville, D. A.; Schowalter, W. R. In Colloidal Dispersions, Cambridge University Press: Cambridge, UK, 1989; Ch. 4–5. 36. Evans, D. F.; Wennerstro¨m, H. The Colloidal Domain, 2nd ed.; Wiley-VCH: New York, 1999.
1146
GUYARD ET AL.
37. Israelachvili, J. Intermolecular and Surfaces Forces, 2nd ed.; Academic Press: London, UK, 1991; Ch. 11. 38. Haw, M. D.; Gillie, M.; Poon, W. C. K. Langmuir 2002, 18, 1626. 39. Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M.; Cosgrove, T.; Vincent, B. Polymers at Interfaces; Chapman & Hall: London, 1993. 40. Suzuki, F.; Nakane, K.; Piao, J.-S. J Mat Sci 1996, 31, 1335–1340.
41. Nakane, K.; Yamashita, T.; Iwakura, K.; Suzuki, F. J Appl Polym Sci 1999, 74, 133–138. 42. Kotoky, T.; Dolui, S. K.; J Sol–Gel Sci Technol 2004, 29, 107–114. 43. This paragraph has been quoted from reviewer’s #2 comments. 44. Schultz, G. V. Z Phys Chem B 1935, 30, 379– 384. 45. Pontoni, D.; Finet, S.; Narayanan, T.; Rennie, A. R. J Chem Phys 2003, 119, 6157–6165.
Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
