Experimental trends in polymer nanocomposites—a review

Materials Science and Engineering A 393 (2005) 1–11

Review

Experimental trends in polymer nanocomposites—a review Jeffrey Jordana , Karl I. Jacobb , Rina Tannenbaumc , Mohammed A. Sharafb , Iwona Jasiukd,∗ a School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0405, USA School of Polymer, Textile and Fiber Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0295, USA c School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0245, USA d Department of Mechanical and Industrial Engineering, Concordia University, Montreal, Que. H3G IM8, Canada

b

Received 17 November 2003; received in revised form 24 September 2004; accepted 27 September 2004

Abstract A review of the recent work on polymer matrix nanocomposites is presented. This review is not intended to be comprehensive, but provides an overview of the processing techniques and trends in the mechanical behavior and morphology of nanocomposites. A number of composite systems with amorphous and/or crystalline polymer matrices and different nano-sized filler materials are discussed. © 2004 Elsevier B.V. All rights reserved. Keywords: Nanocomposites; Polymer matrix composites; Mechanical properties

Contents 1.

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2.

Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2

3.

Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Viscoelastic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Crystallinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Density/volume change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 3 6 6 7

4.

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8

5.

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

1. Introduction Polymer systems are widely used due to their unique attributes: ease of production, light weight, and often ductile

∗ Corresponding author. Tel.: +1 514 848 2424x3143; fax: +1 514 848 3175. E-mail address: [email protected] (I. Jasiuk).

0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2004.09.044

nature. However, polymers have lower modulus and strength as compared to metals and ceramics. One way to improve their mechanical properties is to reinforce polymers with inclusions (fibers, whiskers, platelets, or particles). The embedding of inclusions in a host matrix to make composites, which gives material properties not achieved by either phase alone, has been a common practice for many years. Using this approach, polymer properties can be improved while maintaining their light weight and ductile nature [1–6]. Improvements

2

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

in properties can often be found even at relatively low filler content [4,7,8]. Traditionally, composites were reinforced with micronsized inclusions. Recently, processing techniques have been developed to allow the size of inclusions to go down to nanoscale. For this work, the nano-sized inclusions are defined as those that have at least one dimension in the range 1–100 nm. Experiments have shown that nanoscale reinforcement brings new phenomena, which contribute to material properties. In this review, the interest is in how mechanical properties of polymer matrix composites are altered by introducing nano-sized versus micron-sized reinforcement, and what additional factors contribute to the material response of nanocomposites. Current micromechanics theories rely on the idea that the effective properties of composite materials, such as Young’s modulus, are functions of properties of constituents, volume fraction of components, shape and arrangement of inclusions, and matrix-inclusion interface. These theories, therefore, predict that the properties of composite materials are independent of the size of inclusions. In general, this is correct for systems with micron size reinforcement, but, as mentioned above may not be true for nanocomposite systems. With the recent developments in the nanoscience and nanotechnology fields, the correlation of material properties with filler size has become a point of great interest. As a result, much of the work is still ongoing and there is yet to be a definite conclusion on the effect of nano-sized inclusions on polymer systems. In the following sections, a review is presented on the processing, experimental results, and possible interpretations of those results for polymer matrix nanocomposites.

2. Experimental procedures Making good samples of polymer matrix nanocomposites is a challenging area that draws considerable effort. Researchers have tried a variety of processing techniques [9–16] to make polymer matrix nanocomposites. These include melt mixing, in situ polymerization, and other approaches. Creating one universal technique for making polymer nanocomposites is difficult due to the physical and chemical differences between each system and various types of equipment available to researchers. Each polymer system requires a special set of processing conditions to be formed, based on the processing efficiency and desired product properties. The different processing techniques in general do not yield equivalent results [17]. Vollenberg and Heikens [9] were able to produce good nanocomposite samples by thoroughly mixing filler particles with polymer matrix. The polymer matrices used in these experiments were polystyrene (PS), styrene–acrylonitrile copolymer (SAN), polycarbonate (PC) and polypropylene (PP). The inclusions were alumina beads 35 nm and 400 nm in size and glass beads 4, 30, or 100 ␮m in diameter. The

volume fraction of particles ranged from 0 to 25%. Sample preparation consisted of dissolving polymers in a polar solvent and mixing in the beads for several hours. The mixture was then poured over a large surface to allow the solvent to evaporate and it was subsequently dried under a vacuum at 100 ◦ C. This mixture involved 30% volume fraction of particles. Pure polymer was then added to samples to achieve the desired particle volume fractions. Chan et al. [11] made nanocomposites with polypropylene (PP) matrix and calcium carbonate (CaCO3 ) through melt mixing of the components. First the components were dried in an oven at 120 ◦ C and then cooled to a room temperature. The polypropylene was mixed first with an anti-oxidant. The CaCO3 nanoparticles, which were 44 nm in diameter, were then added slowly and mixing continued for a fixed time after all particles were added. This technique produced good, reasonably well-dispersed samples at lower filler volume fractions, 4.8% and 9.2%, but aggregation was found at a higher volume fraction, 13.2%. Other researchers have also used a form of melt-mixing to produce nanocomposites [17]. Polyurethane–silica nanocomposites have been made by P´etrovic et al. [10] by first mixing the silica with polyol. The mixture was then cured with diisocyanate at 100 ◦ C for 16 h in presence of 0.1% catalyst Cocure 55 (from CasChem) and subsequently poured into a mold. The particles were spherical with an average diameter of 12 nm and had narrow size distribution (10–20 nm). Good samples were produced at 10%, 20%, 30%, and 40% filler weight fraction, but the 50% filler weight fraction sample did not cure completely. It is believed that one of the main issues in preparing good polymer matrix nanocomposite samples is the good dispersion of the nanoparticles in a polymer matrix. Rong et al. [12] grafted monomers made of styrene to surround the particles to produce better dispersion. They used isotactic polypropylene as the matrix and SiO2 as inclusions which were approximately 7 nm in size. The particles were heated first to remove any water absorbed on the surface. Then, they were mixed with one of the monomers and solvent. This mixture was then irradiated and the solvent was removed. Samples were then made by adding polypropylene, tumble mixing, compounding and extruding. This technique produced samples without aggregation and, in addition, greatly increased the particle–polymer matrix interfacial interaction. In situ polymerization is another technique that has been used to make polymer matrix nanocomposites [13–15,18,19]. It is a method in which particles are dispersed first in monomer and then the mixture is polymerized. In Yang et al. [13], nanocomposites with polyamide-6 matrix and silica inclusions were prepared by first drying the particles to remove any water absorbed on the surface. Then, the particles were mixed with ␧-caproamide and concurrently a suitable polymerization initiator was added. The mixture was then polymerized at a high temperature under nitrogen [13]. This technique produced well-dispersed samples when the inclusions were around 50 nm in size, but aggregation occurred for smaller particles around 12 nm in size [15]. This was most

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

likely due to the increased surface energy for the smaller particles, which, in the absence of a stabilizer, favored further particle segregation. Ash et al. [20,21] and Siegel et al. [6] used yet another technique and produced nanocomposites with polymethylmethacrylate (PMMA) matrix and alumina inclusions, which were well dispersed throughout the matrix. The particles were added to methylmethacrylate (MMA) monomer and dispersed through sonication in the low-viscosity solution. An initiator and a chain transfer agent were added later. The mixture was then polymerized under nitrogen, broken into smaller pieces, and dried in a vacuum. Tensile specimens were made through compression molding and cooled to room temperature under pressure. Li et al. [16] used a unique approach to prepare highdensity polyethylene (HDPE)–polypropylene nanocomposites. Seventy-five percent weight HDPE and 25% polypropylene were melt-mixed and extruded into tapes. These tapes were then cut into shorter pieces and melt-processed by either extrusion or compression molding. This produced a nanocomposite with HDPE as the matrix and polypropylene fibrils ranging from 30 to 150 nm in diameter. For clay nanocomposites, the specific choice of processing steps depends on the final morphology required in the composite, i.e., exfoliated or intercalated form. In the intercalated form, matrix polymer molecules are introduced between the ordered layers of clay resulting in an increase in the interlayer spacing, but still maintaining the order. On the other hand, in an exfoliated form, clay layers are separated and are distributed within the matrix. Intercalated nanocomposites are generally formed by melt blending or by in situ polymerization. Exfoliation ability depends on the nature of clay, blending process, and the agents used for curing. The final structure of clay composite has a wide range of variations, depending on the degree of intercalation and exfoliation. A number of experimental techniques were used to characterize intercalated or exfoliated nanocomposites, such as X-ray diffraction, transmission electron microscopy (TEM), DSC, and thermogravimetric analysis (TGA). Thus, in general, there is no single procedure that is followed for making polymer matrix nanocomposites. A combination of melt mixing, extrusion, and compression molding seems to be the method of choice for many researchers. As seen from the above mentioned summary the most important factor to consider in deciding between different processing techniques is the requirement of good particle dispersion in the polymer matrix. This dispersion will play a major role in the results reviewed below.

3. Experimental results While there has been much experimental work done in the area of polymer matrix nanocomposites, there is yet to be a consensus on how nano-sized inclusions affect material properties. This is partly due to the novelty of the area, and

3

the challenges in processing of nanocomposites, lack of systematic experimental results, and scarcity of theoretical studies. Moreover, some material properties have been studied more in-depth than other, leaving gaps in the knowledge on nanocomposite behavior. The following sections will outline some of the experimental results that are available to-date and identify the trends that can be obtained from these results. The focus of this summary will be on particles and matrix morphology (degree of crystallinity, particle size and arrangement) and the mechanical behavior of the nanocomposite materials (elastic modulus, strength, yield stress, strain-tofailure, fracture toughness, and viscoelastic properties). 3.1. Mechanical properties Vollenberg and Heikens [9] performed a wide range of tests on composites with matrices made of polystyrene (PS), styrene–acrylonitrile copolymer (SAN), polycarbonate (PC) and polypropylene (PP) with micro- as well as nano-sized glass or alumina inclusions. The PS and PP were observed to have a very poor interaction with the particles due to their nonpolar character while the interfacial interaction was much more noticeable for the SAN and PC. For composites with PS, PC and PP matrix and glass or alumina particles the modulus increased with decreasing size of particles. In order to investigate the effect of interface adhesion separate experiments were done on SAN and PC matrix composites in which particles were pretreated with siloxane DF 1040 (General Electric Plastics) to decrease the filler–matrix adhesion. No size effect on elastic modulus was found for composites with SAN and PC matrix with either excellent adhesion or very poor adhesion due to treated particles. However, these experiments were done for three different particle sizes which were in micron range only, not nanosize range. For all systems the elastic modulus increased with increasing volume fraction of the inclusions. It is necessary to point out that in these experiments, while there were several different sizes within the micron range, only one particle size can be considered in the nanosize regime by the definition provided earlier. Similar conclusions were obtained by Chan et al. [11] for composites with polypropylene (PP) matrix and calcium carbonate (CaCO3 ) nanoparticles. In their system the CaCO3 inclusions had an average size of 44 nm and a strong interaction with the polymer matrix. The addition of CaCO3 nanoparticles to a PP matrix produced an increase in the elastic modulus compared to the pure matrix as shown in Fig. 1. The increase in modulus coincided with an increase in nanoparticle volume fraction. The reverse effect was found for the yield stress and the tensile strength of the composites; both of these quantities were highest for the pure polypropylene and decreased as the volume fraction of CaCO3 increased. The strain-to-failure did not change much between the various systems. Clay-reinforced nanocomposites have received considerable attention in recent years (more than 100 articles have been published in the literature on clay composites

4

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

Fig. 1. Polypropylene with CaCO3 (from [11]).

in the past three years). A number of polymers, such as PC, PAN, PP, etc. were used as the matrix. Shelley et al. [22] examined a polyamide-6 system with clay platelets. The platelets constituted 2% and 5% weight fraction and were 1 nm × 10 nm × 10 nm in size [23]. Good interaction was found between the matrix and inclusions. With this setup, the elastic modulus was found to improve for both the 2% and 5% samples. For the smaller weight fraction (2%), the increase in effective elastic modulus was 40% over the modulus of the pure polymer system. The larger weight fraction (5%) improved the effective modulus by a factor of two as compared to that of pure polymer. These results were for tensile specimens cut in both longitudinal and transverse directions. In addition, the yield stress also improved for both weight fractions, with the greatest improvement found for the higher concentration of inclusions. The other quantity studied was the strain-to-failure. The 2% system was found to give higher strain-to-failure than the pure system in the longitudinal direction but close to that in the pure system in the transverse direction. The higher filler content resulted in a decline in strain-to-failure from the pure system in both directions [22]. Properties of nanocomposites are highly related to theirs microstructure [24]. For epoxy–silicate clay nanocomposites, Luo and Daniel [24] found that the ideal case for maximizing stiffness and thermal properties is through full exfoliation and dispersion, which is not achieved usually, but often one obtains partial exfoliation and intercalation. Similar results were presented by Mark and co-workers [25] for clay reinforced natural and epoxidized rubber. For sodium montmorillonite silicate clay the degree of its dispersion determined its reinforcing effect in rubber. Zhang et al. [26] further demonstrated the similar behavior for clay–polypropylene nanocomposites. They also showed that the reinforcing effects demonstrated by storage modulus, thermal stability, etc., are directly related to the dispersion of clay particles in the matrix. Park et al. [27] fabricated clay nanocomposites where organophilic clay particles were embedded in a syndiotactic polystyrene with melt intercalation. They found that stepwise mixing method (where styrene polymer is blended with the clay first to enable intercalation followed by blending with polystyrene matrix) resulted in a higher degree of intercalation (than a simultaneous mixing method) and that

the higher degree of intercalation in the structure resulted in higher tensile strength and other mechanical properties. In another work [28] on polystyrene–clay nanocomposites, prepared by free-radical polymerization, the authors found that exfoliated nanocomposites had better thermal stability and better mechanical properties than pure polystyrene. With clay–polyurethane nanocomposites, Tortora et al. [29] showed that exfoliation occurred for low montmorillonite content but when the clay content was increased more intercalation was observed associated with higher elastic modulus and yield strength. However, breaking stress and breaking strain decreased with increasing clay content. Masenelli-Varlot et. al [30] studied the effects of polyamide-6 with intercalated and exfoliated montmorillonite. The intercalated samples ranged from 1.7 to 2.7% clay while the exfoliated samples ranged from 2.4 to 4.2%. Under low-deformation tests, Young’s modulus increased linearly with filler content with no major difference between intercalated and exfoliated. Under compression tests, the exfoliated samples had one direction that was higher than the rest while the intercalated samples did not. The intercalated samples are likely weaker due to the weaker mechanical coupling. The ultimate stress and strain-to-failure were lower for all filler concentrations than pure polyamide-6. With the same level of dispersion and filler content, the exfoliated samples exhibited greater stress at failure than the intercalated samples while the intercalated samples had a greater strain-to-failure [30]. Few micromechanical models were developed which are in agreement with the behavior of some nanocomposites [31–35]. A recent theoretical study [34] shows that when the structure changes from intercalated to an exfoliated structure, the morphological change is accompanied by a moderate change in modulus rather than an abrupt change. This theoretical prediction was supported by experimental results in [36]. There is a general agreement in the literature that exfoliated systems have better mechanical properties, particularly higher modulus, than intercalated nanocomposites [37]. Also, studies on nylon 6–clay nanocomposites showed that morphology or physical properties are not significantly reduced from reactions or polymer molecular degradation during the processing stage [38]. Polyamide-6 nanocomposites with silica inclusions have been examined (e.g. [15]). In contrast to the polyamide–clay system described in [22], the nanoparticles were 17, 30, or 80 nm in diameter. The elastic modulus was slightly higher for the nanocomposites than for the pure system, but there was little difference between composites with different nanoparticle sizes. This can be seen in Fig. 2 where S refers to the smallest particle size and L to the largest particle size. As in the polyamide–clay system, the yield stress increased with increasing filler content and also increased slightly with the decreased size of the particles. The strain-to-failure followed the opposite pattern as it decreased greatly for an increase in volume fraction and decrease in particle size. In addition, there existed a sharp decrease in the stress immediately before failure in all systems [15]. A separate study involving

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

5

Fig. 3. Stress strain curve for PMMA and alumina (from [20]).

Fig. 2. Polyamide-6 with small (S) and large (L) inclusions (from [15]).

these same materials examined the influence of a silane pretreatment, which improved the matrix–filler interface. For the treated system, there was an improvement in strength as well as toughness, which was in contrast with the untreated system in which strength improved as toughness decreased [19]. Another study centered on polyurethane–silica composites [10]. Silica nanoparticles have relatively strong interaction with polyurethane matrix. Polyurethane–silica composites were formed with inclusions of 12 nm and 1.4 ␮m at 10–50% weight fraction of inclusions. The tensile strength varied little between the micron and nano-sized particles, up to a filler composition of 20% weight fraction. Above 20%, the tensile strength was higher for the nanocomposite as compared to pure polyurethane polymer. In the nanocomposite, the strain-to-failure increased by 500% over the pure system while for composites with larger inclusions the increase was only by 100% [10]. A unique set of experiments was performed by Rong et al. [12] with SiO2 nanoparticles in a polypropylene matrix with and without the addition of grafting polymers (either PS or PMMA), which have the ability to improve dispersion of nanoparticles and interfacial interactions between the particles and the matrix. The elastic modulus increased with increasing volume fraction in both systems but was much higher in the untreated composite systems (without grafting polymers). The modulus for the composites treated with grafting polymer only improved minimally as a function of volume fraction of SiO2 particles. The opposite effect was found for the tensile strength where the treated composites showed increased strength while the strength changed very little in the untreated composites as a function of volume fraction of particles. Interestingly, the results for strength were insensitive to change in volume fraction of particles. The strain-to-failure was lower in the untreated systems than the pure polypropylene samples. In the treated system the strain-to-failure increased for volume fraction of particles up to 2% and then started decreasing. This change in behavior at approximately 2% volume fraction was attributed to the change in disper-

sion of particles, namely the change from the uniform dispersion to clustering resulting in microcomposite agglomeration [12]. In another study on palladium–PMMA nanocomposites showed that addition of Pd in PMMA increased the oxidative thermal stability of the system. Improvement in thermal stability was linearly related to the volume fraction of palladium inclusions [35], although the elastic modulus decreased as a function of the Pd content. Very interesting results were obtained by Ash et al. [20] who performed a series of tests on PMMA–alumina nanocomposites. The alumina nanoparticles were around 40 nm and had very little interfacial interaction with the polymer matrix. As filler content increased, there was a sharp initial drop in Young’s modulus followed by a steady increase. Even at the highest filler content, however, the effective elastic modulus was lower than the pure system. In addition, the yield stress and tensile strength of the pure matrix was higher than for the composite as shown in Fig. 3. These results for the strength are very different from the other systems that have been examined. In addition, the strain-to-failure increased by around 800% over that for the pure system. Elastic properties were found to increase for a nanocomposite system made 75 wt% of high-density polyethelyne (HDPE) matrix and 25 wt% polypropylene (PP) [16]. During processing the compound was fibrillated and PP nanofibers were created. The fibers were cylindrical in shape, 30–150 nm in diameter. The elastic modulus, yield stress, and tensile strength were greater for the nanocomposite system than for samples of pure HDPE [16]. However, the strain-to-failure was lower for nanocomposite systems. From the above discussion, it is possible to extract a few trends for the behavior of polymer matrix nanocomposites based on the nature of the polymer matrix, particularly crystalline or amorphous nature of the polymer, and the interaction between the filler and matrix. The elastic modulus tends to increase with the volume fraction of inclusions in every case. In some systems, there is a critical volume fraction at which aggregation occurs and the modulus goes down [1–4,8]. In general, there is also an increase in modulus as the size of the particle decreases. Interaction between matrix and filler may play an important role in the effects of the nanoparticles [1–3] on composite properties. For polymer systems capable of having a higher degree of crystallinity, the increase in modulus with decreasing particle size is found to

6

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

be greater in systems with poor interaction between filler and matrix as opposed to those with good interaction. However, the overall trend of the modulus of polymer nanocomposites is not found to be greatly dependent upon the nature of the matrix nor the interaction between filler and matrix. An examination of the yield stress gives a different trend than that of the elastic modulus. For composites with good interaction between filler and matrix, the yield stress tends to increase with increasing volume fraction and decreasing particle size, similarly to the increase in modulus under same conditions. The pattern changes when there is poor interaction between the matrix and particles. The addition of nanoparticles with poor interaction with the matrix causes the yield stress to decrease, compared to the neat matrix, regardless of the filler concentration or size. The ultimate stress follows a similar pattern as that observed for the yield stress. It generally increases in polymer systems (both crystalline and amorphous) with good filler–matrix interaction and it increases in general as particle size decreases. There is no uniform trend with respect to the volume fraction of particles for the ultimate stress. A poor filler–matrix interaction leads to a decrease in the ultimate and yield stress as compared to the pure matrix system. The change in strain-to-failure behavior for nanocomposites, like the yield and ultimate stress, is different depending on the system. In this case, however, the change is with respect to the nature of the polymer matrix as opposed to the filler–matrix interaction. In general, the addition of nanoparticles to a mostly crystalline or semi-crystalline polymer, regardless of the filler–matrix interaction, reduces the maximum strain. The opposite trend is found in amorphous polymers with the increase in strain-to-failure coinciding with a decrease in particle size. 3.2. Viscoelastic properties Viscoelastic properties were found to improve with the addition of particles in a polyamide-6 matrix with clay platelet inclusions. More specifically, the storage modulus was found to be higher for the composite systems than for the pure polyamide-6 system. The increase in modulus coincided with the increase in filler content [22,30]. However, experimental results also showed that the onset of the decrease in modulus occurred at lower temperatures as the filler content increased [22]. The storage modulus was also found to increase with increasing volume fraction for polyamide-6 nanocomposite with silica nanoparticles rather than clay platelets [15]. A similar increase in storage modulus was found for poly(vinylidene fluoride)–clay nanocomposites [39]. Viscoelastic properties were also found to improve in other systems, including polyurethane–silica composites. The storage modulus increased for such composites with both micron and nano-sized particles. This modulus also increased as the weight fraction of the particles was increased to 50%. It is also noteworthy to mention that in the rubbery state of the system, the modulus is highest for the nanocomposite when the

weight fraction of the filler is as large as 40 or 50%. However, if the weight fraction is below this level, the microcomposite has the higher modulus [10]. In general, viscoelastic properties tend to be higher in nanocomposites than in pure polymer systems. When there is good filler–matrix interaction, the storage modulus generally increases with increasing volume fraction. The modulus also seems to increase as the particle size decreases. However, there is little experimental work in this area for composites with poor filler–matrix interactions. Overall, the storage modulus tends to increase with the presence of nanoparticles in a composite system. Morphological details, such as exfoliation, intercalation, or cross-linked matrix versus uncross-linked matrix, have a significant effect on the viscoelastic properties of nanocomposites. In epoxy–clay nanocomposite system forces from cross-linked epoxy molecules cause intercalated clay galleries to exfoliate, associated with a gradual increase in viscosity and a relatively fast increase in storage modulus [40]. In maleated polypropylene–layered silicate nanocomposites, exfoliated nanocomposite systems were reported to have higher shear and complex viscosities, while in a shear flow exhibiting the largest drop in complex viscosity due to alignment of clay layers [41]. Dynamic storage moduli also show a similar behavior [41]. Experiments with clay–nylon12 nanocomposite systems have shown that melt processing under low shear improves intercalation, while larger shear enhances exfoliation associated with a decrease in melt viscosity [42].

3.3. Crystallinity The degree of crystallinity of a polypropylene system was found to change very little with the addition of CaCO3 nanoparticles. On the other hand, the peak crystallization temperature was increased by around 10 ◦ C with the addition of the CaCO3 inclusions. This increase was found to be nearly the same for all filler volume fractions. Moreover, the size of the crystalline domain spherulites was found to change dramatically in different systems. Scanning electron microscopy (SEM) showed that the spherulites in the pure system had an average size of around 40 ␮m, but no spherulites could be seen when CaCO3 nanoparticles were added to the system [11]. It is possible that the spherulites were too small and they were not detected by SEM. The addition of nano-sized montmorillonite clay platelets to a polyamide-6 matrix was found to have no effect on the degree of crystallinity of the polymer with weight fraction up to 5%. However, the nanoparticles did have an effect on the size of the crystallites. The size of crystallites in the nanocomposites was nearly an order of magnitude smaller than the size of spherulites in the pure matrix system [22]. Results for a polyamide-6 matrix composite with silica nanoparticles showed that neither the size nor the filler content had any effect on the crystallinity of the system [15].

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

7

Fig. 4. Young’s modulus and glass transition temperature vs. filler weight fraction (from [20,21]).

In a separate experiment with clay–polyamide nanocomposites, the glass transition temperature of the system increased with increasing clay nanoparticle content. This increase was attributed to either the effect of the clay layers retarding the polyamide molecule main-chain motion or a slightly higher molecular weight of the polymer due to the reactive organoclay used [43]. Another study, however, found a 10 ◦ C drop in the glass transition temperature for all clay filler contents. This drop was attributed to the presence of surfactants [30]. Ash et al. [21] studied the glass transition temperature of PMMA–alumina nanocomposites. They found that for filler weight fractions less than 0.5%, the glass transition temperature did not change, but for filler weight fraction above 0.5% the glass transition temperature dropped by more than 20 ◦ C and stabilized at a new temperature at weight fractions above 1% as shown in Fig. 4b. This drop occurred at the same weight fraction as the drop in the elastic modulus, as shown in Fig. 4a, and as mentioned in previous discussion about this system. A different result was found for the glass transition temperature of a polyurethane matrix with silica inclusions. In this nanocomposite system, the glass transition temperature was found to increase as filler concentration increased. Different techniques produced differing results but the consensus was that the glass transition temperature increased by around 10 ◦ C for the nanocomposite above the neat resin as the weight fraction of the filler was increased to 50%. The transition temperature also increased with the use of micronsized inclusions, but the increase was not as great as that for the nanocomposite [10]. The pattern that the crystallinity was little affected by the addition of particles was also observed in a polypropylene–SiO2 composite that was treated with grafting polymers. The particles were found to have a small nucleation effect on the polypropylene matrix, but, overall, the crystallinity was not greatly influenced by the addition of the nanoparticles [12]. Park et al. [37] found that for (syndiotactic) polystyrene–organoclay nanocomposites, dispersed clay layers act as nucleating agent competing with crystal growth in polymers. Thus, exfoliated clay nanocomposites have lower degree of crystallinity with faster crystallization rate. However, the crystallinity of crystalline and semi-crystalline

polymers is not affected very much by the addition of nanoparticles. There may be some changes in particular nanocomposite systems, but overall no major differences in crystallinity of nanocomposites versus neat polymers were observed in any of the systems examined. On the other hand, the glass transition temperature was influenced by the addition of particles. When there is good filler–particle interaction, the glass transition temperature tends to increase with a decrease in the size of particles for amorphous polymers. For crystalline polymers, the transition temperature decreases with an increase in particle concentration. For an amorphous system with poor filler–polymer interfacial interaction, the glass transition temperature decreased overall. Thus, while the degree of crystallinity is not significantly affected by the presence of particles, the glass transition temperature is very dependent upon this factor. 3.4. Density/volume change It is natural for the density of a composite to be higher than the density for a pure polymer system. The inclusions typically have much higher density than the polymer leading to a higher density for the composite. This was found to be true in a polyurethane matrix with micron and nano-sized silica inclusions. The density increased, as expected, as the filler content increased. However, the density of the composites with the micron-sized inclusions was higher than the density for the nano-sized inclusions [10]. The volume change of a composite undergoing tensile elongation was examined by Reynaud et al. [15] using a polyamide-6 composite with silica inclusions of various sizes within the nanometer range. Reynaud et al. [15] reported a volume increase for the pure polyamide-6 but larger increases for the composite systems. The volume change was highest for composites containing the smallest particles, which provides support for their debonding model that will be discussed later in this article. Investigations examining the change in the density or volume of polymer nanocomposites are limited. Nevertheless, the trend is that nano-sized inclusions affect the volume of a composite more than micron-sized inclusions do. The volume for the composite is higher for a nanocomposite than a microcomposite with the same filler weight and the same

8

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

polymer weight. It is important to point out that these results are only for composites with relatively good filler–polymer interaction.

4. Discussion From the experimental results presented above, we observe that different composite systems can lead to very different results. One important observation is that composites with nano-sized inclusions generally have different properties than composites with larger scale inclusions. The specific reasons why the polymer matrix composites with nano-sized reinforcement have different properties than composites with micron-sized reinforcement are not fully understood, but several theories have been introduced to explain some of the changes in material morphology and behavior that are seen at the nano-scale. It is important to point out, however, that most of these theories were developed to explain particular results and, therefore, are not necessarily applicable to a large number of polymer nanocomposites. Chan et al. [11] proposed that properties such as elastic modulus, tensile strength, and yield strength decrease in nanocomposites with polypropylene matrix due to the change in nucleation caused by the nanoparticles (Fig. 5). The nanoparticles produce a much larger number of nucleating sites but, in turn, greatly reduce the size of these spherulites. In their experimental work, no spherulites were found in the nanocomposites by SEM indicating that either none were present or they were reduced to such a small size that SEM could not detect them. It was further proposed that there was another mechanism which was causing these same properties to increase. The increase occurred when there was a strong interaction between the polymer and filler. This interaction had larger impact in nanocomposites due to the large interfacial area between the filler particles and the matrix. Other investigators have suggested that this interaction leads to a layer of polymer that is directly adsorbed and bound to the particles [9,10,20,21]. Experimental work that has been performed on a polystyrene–cobalt nanocomposite with cobalt nanoparticles with an average size of 21 nm has shown that the polymer layer was about 24 nm, and varied non-linearly with molecular weight [44].

An increase in yield and tensile strength and modulus in nanocomposite systems as compared to microcomposites can be partially explained on the basis of the interaction between the filler and the matrix. It has been found that a greater adhesion between the matrix and inclusion causes less debonding when a stress is applied and, consequently the elastic modulus and strength are improved [45]. Vollenberg and Heikens [9] explained that if there is a strong interaction between the polymer and the particle, the polymer layer in the immediate proximity of the particle will have a higher density. For most systems, density is proportional to elastic modulus, so the region directly surrounding the inclusions will be a region of high modulus. The polymer right outside this high modulus region will have a lower density due to the polymer chains that are moved towards the particle. For large particles, the size of the low density region will be relatively large, and the contribution of the high modulus filler will be diminished. For nanoparticles, the number of particles for a given volume fraction is much larger, thus the particles will be much closer to one another. If the particles are densely packed, then the boundary layer of polymer at the interface will comprise a large percentage of the matrix and can create a system where there is no space for a low modulus region to form. This results in the elastic moduli of composites with smaller particle size (nano) being greater than the moduli of composites with larger inclusions [9,10]. The small interparticle distance in nanocomposites was used as another parameter to explain the changes in the elastic modulus and strength of these materials when compared with the composites with micron-sized particles. The same parameter also plays a role in the glass transition temperature changes observed in nanocomposites versus composites with micron-sized reinforcement. Ash et al. [21] found that for their system the glass transition temperature was constant until around 0.5% weight fraction of particles, then had a sharp drop, and then it remained constant for weight fractions above 1%. When there is little or no interfacial interaction between the filler and matrix and the interparticle distance is small enough, the polymer between two particles acts as a thin film. For a thin film, the glass transition temperature decreases as film thickness decreases. The distance between particles in a composite with the filler weight fraction below 0.5% is relatively large, and hence, in this case the

Fig. 5. (a) Pure polypropylene and (b) polypropylene with 9.2% volume filler (from [11]).

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

Fig. 6. Debonding around 50 and 12 nm particles (from [15]).

polymer between each particle is not considered to belong to the thin film regime. As the filler concentration increases, the interparticle distance and the resulting thickness of the film, decrease. This theory, however, does not explain why the glass transition temperature levels off rather than continues to drop as a function of increasing weight fraction of the filler. A drop-off in Young’s modulus was found for the same filler weight fraction as the drop in glass transition temperature. It was proposed that as the glass transition temperature decreased, the relative testing temperature increased. Also, the elastic modulus of the matrix, PMMA, decreases as tem-

9

perature increases, so the drop in glass transition temperature is correlated with the drop in modulus [20,21]. Reynaud et al. [15] found that during tensile testing, the volume of polymer nanocomposites increased, with the greatest increase occurring in systems with the smallest particles. To explain this, the debonding process of the polymer next to the inclusions was examined, as shown in Fig. 6. It was proposed that the smallest particles tend to aggregate and debonding occurs around each individual particle. As a result, the large clusters of small particles act as larger soft particles. On the other hand, the larger filler particles do not aggregate and each particle undergoes a single debonding process. Due to the different results obtained and the different nature of the various polymer nanocomposite systems, there is no observed universal trend that can be modeled and explained. There are, however, observations that show the behavior of nanocomposites different from composites with larger scale inclusions. The particle size and the polymer and particle morphology tend to play a very important role. In addition, the nature of dispersion and aggregation of particles can affect the properties of composites significantly. Filler–matrix interaction is another factor that influences the properties. The strength of the interaction plays a role in the thickness and density of the interphase, which consists of a layer of high density polymer around the particle. The effects of the interface on the behavior of a composite depend upon the interparticle distance. For constant filler content, with reduction in particle size, number of filler particles in-

Table 1 Summary of polymer nanocomposite trends Crystalline

Amorphous

Increase w/volume fraction Increase or no change with decrease of size Increase w/volume fraction Increase w/decrease size Greater increase than for good interaction

Increase w/volume fraction Increase w/decrease size Increase w/volume fraction Increase w/decrease size

Good interaction

Increase w/volume fraction Increase w/decrease size Decrease with addition of particles

N/A

Good Interaction

Decrease with addition of particles

Poor interaction

Ultimate stress/strain

Increase w/decrease size No unified result for change inVf Lower than pure for small volume fractions

Nano>micro after 20%weight

Good interaction

Decrease with addition of particles

Poor interaction

Density/volume

Increased volume as size decreases N/A

Increased volume as size decreases N/A

Good interaction Poor interaction

Strain-to-failure

Decrease with addition of particles

Good interaction

Decrease with addition of particles

Increase with addition of particles Increase w/decrease size Increase with addition of particles

Tg

Decrease with addition of particles N/A

Increase w/decrease size Level until 0.5%, drops off level from 1–10%

Good interaction Poor interaction

Crystallinity

No major effect No major effect

N/A N/A

Good interaction Poor interaction

Viscoelastic

Increase w/volume fraction Increase w/decrease size N/A

Increase w/volume fraction nano less regular

Good interaction

Decrease with addition of particles—drop at 1% with rise following

Poor interaction

Elastic modulus

Yield stress/strain

Poor interaction

Poor interaction

10

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11

creases, bringing the particles closer to one another. Thus, the interface layers from adjacent particles overlap, altering the bulk properties significantly. These issues play a major role in the effect of nano-sized inclusions in a polymer matrix. For nanoparticles, any configuration changes in the matrix will have a significant effect when the characteristic radius of polymer chains is of the same order as the inclusions [45,46]. There are other areas addressed in literature. For example, L´opez et al. [47] examined the processing and thermal and mechanical properties of magnetic nanocomposites. In another work the mechanical properties of clay nanocomposites were analyzed as a function of filler loading and orientation [30]. Zhang et al. [48] provided a look at matrix–filler interfacial properties. Effect of matrix on the polymer matrix composites were examined by Friedlander et al. [49].

5. Conclusions Currently, a significant amount of work is being published on polymer nanocomposites. In an attempt to further understand the synthesis, processing, and properties analysis of polymer nanocomposites, a selection of representative and recent literature was chosen to highlight some of the issues related to the preparation and mechanical behavior of composites with nano-sized reinforcement in comparison with composites with larger micron-sized inclusions. As was discussed above, some trends are observed but no universal patterns for the behavior of polymer nanocomposites can be deduced in general. These observed trends are summarized in Table 1. In general, the material properties of polymer nanocomposites are superior to the pure polymer matrix or composites with larger sized inclusions. The effects of the nanoparticles are dependent on many variables but especially upon the relative crystalline or amorphous nature of the polymer matrix as well as the interaction between the filler and matrix.

Acknowledgments I.J. acknowledges the support by the National Science Foundation (Grant CMS-0085137) and the Air Force Office of Scientific Research (Dr. Thomas Hahn, monitor).

References [1] H. Akita, T. Hattori, J. Polym. Sci. B: Polym. Phys. 37 (1999) 189–197. [2] H. Akita, H. Kobayashi, J. Polym. Sci. B: Polym. Phys. 37 (1999) 209–218. [3] H. Akita, H. Kobayashi, T. Hattori, K. Kagawa, J. Polym. Sci. B: Polym. Phys. 37 (1999) 199–207. [4] J.-H. Chang, Y.U. An, J. Polym. Sci. B: Polym. Phys. 40 (2002) 670–677. [5] S.A. Zavyalov, A.N. Pivkina, J. Schoonman, Solid State Ionics 147 (2002) 415–419.

[6] R.W. Siegel, S.K. Chang, B.J. Ash, J. Stone, P.M. Ajayan, R.W. Doremus, L.S. Schadler, Scr. Mater. 44 (2001) 2061–2064. [7] C.L. Wu, M.Q. Zhang, M.Z. Rong, K. Friedrich, Compos. Sci. Technol. 62 (2002) 1327–1340. [8] R. Magaraphan, W. Lilayuthalert, A. Sirivat, J.W. Schwank, Compos. Sci. Tech. 61 (2001) 1253–1264. [9] P.H.T. Vollenberg, D. Heikens, Polymer 30 (1989) 1656–1662. [10] Z.S. P´etrovic, I. Javni, A. Waddon, G. B´anhegyi, J. Appl. Polym. Sci. 76 (2000) 133–151. [11] C.-M. Chan, J. Wu, J.-X. Li, Y.-K. Cheung, Polymer 43 (2002) 2981–2992. [12] M.Z. Rong, M.Q. Zhang, Y.X. Zheng, H.M. Zeng, R. Walter, K. Friedrich, Polymer 42 (2001) 167–183. [13] F. Yang, Y. Ou, Z. Yu, J. Appl. Polym. Sci. 69 (1998) 355–361. [14] Y. Ou, F. Yang, Z.-Z. Yu, J. Polym. Sci. B: Polym. Phys. 36 (1998) 789–795. [15] E. Reynaud, T. Jouen, C. Gautheir, G. Vigier, Polymer 42 (2001) 8759–8768. [16] J.-X. Li, J. Wu, C.M. Chan, Polymer 41 (2000) 6935–6937. [17] C.I. Park, O.O. Park, J.G. Lim, H.J. Kim, Polymer 42 (2001) 7465–7475. [18] C. Zeng, L.J. Lee, Macromolecules 34 (2001) 4098–4103. [19] Y. Li, J. Yu, Z.-X. Guo, J. Appl. Polym. Sci. 84 (2002) 827–834. [20] B.J. Ash, J. Stone, D.F. Rogers, L.S. Schadler, R.W. Siegel, B.C. Benicewicz, T. Apple, Mater. Res. Soc. Symp. Proc. 661 (2000). [21] B.J. Ash, L.S. Schadler, R.W. Siegel, Mater. Lett. 55 (2002) 83–87. [22] J.S. Shelley, P.T. Mather, K.L. DeVries, Polymer 42 (2001) 5849–5858. [23] A. Okada, T. Usuki, O. Kurauchi, Kamigaito, in: J.E. Mark, C.Y.-C. Lee, P.A. Branconi (Eds.), ACS Symposium, 1995, pp. 55–65. [24] J.-J. Luo, I.M. Daniel, Compos. Sci. Technol. 63 (2003) 1607–1616. [25] Y.T. Vu, J.E. Mark, L.H. Pham, M. Engelhardt, J. Appl. Polym. Sci. 82 (2001) 1392–1403. [26] Y.-Q. Zhang, L.-H. Lee, H.-J. Jang, C.-W. Nah, Compos. Part B: Eng. 35 (2004) 133–138. [27] C.I.L. Park, O.O. Park, J.G. Lim, H.J. Kim, Polymer 42 (2001) 7465–7475. [28] C.-R. Tseng, J.-Y. Wu, H.-Y. Lee, F.-C. Chang, J. Appl. Polym. Sci. 85 (2002) 1370–1377. [29] M. Tortora, G. Gorrasi, V. Vittoria, G. Galli, S. Ritrovati, E. Chiellini, Polymer 43 (2002) 6147–6157. [30] K. Masenelli-Varlot, E. Reynaud, G. Vigier, J. Varlet, J. Polym. Sci. B: Polym. Phys. 40 (2002) 272–283. [31] P. Uribe-Arocha, C. Mehler, J.E. Puskas, V. Altstat, Polymer 44 (2003) 2441–2446. [32] T. Tzianetopoulou, M.C. Boyce, Materials Research Society Symposium Proceedings, vol. 788, Continuous Nanophase Nanostruct. Mater. (2003) 601–607. [33] I.M. Daniel, H. Miyagawa, E.E. Gdoutos, J.J. Luo, Exp. Mech. 43 (2003) 348–354. [34] N. Sheng, M.C. Boyce, D.M. Parks, G.C. Rutledge, J.I. Abes, R.E. Cohen, Polymer 45 (2004) 487–506. [35] C. Aymonier, D. Bortzmeyer, R. Thomann, R. Mulhaupt, Chem. Mater. 15 (2003) 4874–4878. [36] C.-R. Tseng, J.-Y. Wu, H.-Y. Lee, F.-C. Chang, J. Appl. Polym. Sci. 85 (2002) 1370–1377. [37] C.I.L. Park, W.W. Choi, M.H.O. Kim, O.O.K. Park, J. Polym. Sci. B: Polym. Phys. 42 (2004) 1685–1693. [38] T.D. Fornes, P.J. Yoon, D.R. Paul, Polymer 44 (2003) 7545– 7556. [39] L. Priya, J.P. Jog, J. Polym. Sci. B: Polym. Phys. 40 (2002) 1682–1689. [40] J.H. Park, S.C. Jana, Macromolecules 36 (2003) 2758–2768. [41] C.M. Koo, M.J. Kim, H. Min, S.O. Kim, I.J. Chung, J. Appl. Polym. Sci. 88 (2003) 1526–1535. [42] C.Y. Lew, W.R. Murphy, G.M. McNally, Dev. Chem. Eng. Miner. Proc. 12 (2004) 149–158.

J. Jordan et al. / Materials Science and Engineering A 393 (2005) 1–11 [43] H.-L. Tyan, K.-H. Wei, T.-E. Hsieh, J. Polym. Sci. B: Polym. Phys. 38 (2000) 2873–2878. [44] E. Tadd, A. Zeno, M. Zubris, N. Dan, R. Tannenbaum, Macromolecules 36 (2003) 6497–6502. [45] K.H. Wang, I.J. Chung, M.C. Jang, J.K. Keum, H.H. Song, Macromolecules 35 (2002) 5529–5535. [46] A. Ranade, N.A. D’Souza, B. Gnade, Polymer 43 (2002) 3759–3766.

View publication stats

11

[47] D. L´opez, I. Cendoya, F. Torres, J. Tejada, C. Mijangos, J. Appl. Polym. Sci. 82 (2001) 3215–3222. [48] Y. Zhang, S. Ge, B. Tang, M.H. Rafailovich, J.C. Sokolov, D.G. Peiffer, Z. Li, A.J. Dias, K.O. McElrath, S.K. Satija, M.Y. Lin, D. Nguyen, Langmuir 17 (2001) 4437–4442. [49] S.K. Friedlander, K. Ogawa, M. Ullmann, J. Polym. Sci. B: Polym. Phys. 38 (2000) 2658–2665.