The tribology characteristics of the PEEK composites filled with nanoparticles

It has been claimed [54] that the wear resistance of PEEK composites filled with larger ZrO2 nanoparticles measuring 86 nm became worse than that of the unfilled PEEK because of the discontinuous thick transfer film and the weak mutual adhesion. In contrast, the addition of much finer ZrO2 measuring 10 nm could form a thin, uniform and tenacious transfer film on the counterpart steel surface during the wearing process, leading to a lower frictional coefficient and wear rate of the filled PEEK. It seems that the smaller fillers are more effective in increasing the hardness and lowering the wear rate, as also observed or expected in the current 15 nm SiO2 composites. For the present PEEK composites containing both 15 and 30 nm nano particles and both exhibiting appreciable hardness increment, it is conceivable to expect satisfactory wear improvement in composites filled with 7.5 to 10 wt%

nanoparticles.

4.6 The effect of inorganic nano fillers on the tensile properties of PEEK

As shown in Fig. 3.26(a), the elastic modulus of the nanoparticles filled PEEK composites would increase linearly with the content of the filler, but the UTS would not. It was reported that the elastic modulus of the resulting composites, no matter what the matrix and the filler were, would enhance linearly with the content of the inorganic fillers [129]. In the present study, the same trend in the improvement of modulus is also found.

Improving the performance of polymer products by incorporating inorganic fillers has long been an important industrial activity, and traditionally this has been achieved by using materials such as carbon blacks, clays, talc and silica. More recently, the modification of polymer composites using nano-scaled fillers, with their high surface to volume ratios, has

been of increasing interest. It is meaningful to examine the reasoning for the linear improvement in elastic modulus. The following reports provide some insights.

In an extensive study, Tsagaropoulos and Eisenberg [159,160] reported that a range of vinyl polymers exhibit an additional maximum in tanδ about 50^oC above the main a relaxation occurring at the glass transition Tg; and similar behavior has been observed also for poly (dimethylsiloxane) [172]. This additional relaxation was described by Tsagaropoulos and Eisenberg as a ‘second glass transition’ and the model they proposed to account for its presence, envisaged three regions around a nanoparticle; an inner tightly bound layer in which polymer motion is severely restricted by interactions with the surface, an intermediate but more loosely bound layer, and finally the unrestricted bulk polymer. Such a 3-layer model is supported by the NMR data [173], however neutron scattering experiments show only two relaxation times in filled systems [174,175]: a slow process, corresponding presumably to restricted mobility adjacent to the filler surface, and that of the bulk polymer which retains normal segmental dynamics.

According to the proposed model above, polymer reinforced with inorganic fillers would become more rigid close to the interface of the fillers. Accordingly, the resulting polymer nanocomposites would possess higher modulus when the fillers are incorporated. Te mean distance L between the statistically distributed nanoparticles can be roughly estimated by the equation [176],

] 1 ) /

[( −

=d F V_f

L , (4.7)

where F is packing factor, 0.64 for mondispersed sphere, and Vf is volume fraction. Applying

the data in Table 2.1 and the above equation, it is possible to estimate the mean distance L between statistically distributed nanoparticles, as shown in Table 4.8. According to Table 4.8, the mean distance between statistically distributed nanoparticles for the 15 nm silica is apparently less than those of the 30 nm silica and 30 nm alumina. However, as shown in Fig.

3.26, the silica (15 nm) filled PEEK composites possess lower modulus and higher failure strain than those of silica (30 nm) and alumina (30 nm).

The value of the modulus associated with the composites filled with 15 nm silica is thought at first to be higher than those of the other two counterparts. But, a reverse effect is observed. Recalling the model proposed by Tsagaropoulos and Eisenberg, the thickness of the immobilized layer existing adjacent to the filler surface was estimated to be about 1.5 nm for silica-filled poly(dimethylsiloxane) [175] , and to be ~5 nm for polybutadiene [177].

Therefore, the main factors attributing the reverse effect in 15 nm silica might be:

1. The thickness of the immobilized layer is too minor, as compared with the large spacing between the nanoparticles, to play its role into effect.

2. The smaller the nanoparticles the easer to form agglomeration could be, and this effect could detrimentally and significantly lower the contribution of inter-filler distance.

3. The agglomeration in silica (15 nm) might be locally occurred; hence, the failure strains in silica (15 nm) filled PEEK composites are still higher than those of the other two.

In summary, the current volume fraction of the nanoparticles (less than 5 vol%) might still be too low to justify the above argument, since the interspacing between the nanopartilces is appreciably greater than the particle size. Thus the hardening effect from the thin interface mantle layer around the nanoparticles might be overshadowed by the clustering artifact.

4.7 The effect of inorganic fillers on the crystallization of PEEK molecular chains

As mentioned above, the particle content level at 5 wt% would show a local extrema in crystallinity, and this phenomenon could account for the UTS extrema for the filler contents around 5.0 to 7.5 wt%. As the particle contents are further raised to 7.5 or 10.0 wt%, the Xc

values will decrease again to the range of 25 to 30%. It is supposed that the dispersion of the nano-size particles in the PEEK matrix by means of compression molding would result in more or less aggregation in micrometer size. In the interface between PEEK and nanoparticle, a “melt-induced diffusion” phenomenon might occur, and this phenomenon could force the nanoparticles to diffuse into the PEEK matrix, as shown in the TEM images in Figs.

3.28-3.31. The TEM micrographs show that both the isolated nanoparticles and clustering of 5-9 nanoparticles could exist in the PEEK matrix simultaneously. On the other hand, due to the large size difference in the PEEK powders (~100 µm) and the nanoparticles (15 or 30 nm) used in the hot pressing fabrication, there are only partial nanoparticles diffusing into the center of the PEEK powders during melting; the remaining nanoparticles would still be retained in the surrounding area around the micro-sized PEEK powder during melting. It follows that not all nanoparticles could contribute their effects in enhancing heterogeneous nucleation during crystallization, especially for the case with high nanoparticle contents.

The crystallinity was found to be 39% for the pure PEEK at a cooling rate of 2.5 ^oC/min, as shown in Tables 3.7 to 3.9. However, irrespective of the content of particles, the kind of particles, or the size of the particles under investigation, the crystallinities of silica or alumina filled PEEK nanocomposites were found to have the Xc values ranging from 34-40%

at the cooling rate of 2.5 ^oC/min. It was reported [139] that a small amount (0.3-0.9 wt%) of 7 nm silica could increase the crystallinities from 22% for the pristine PEN (poly(ethylene 2,6-naphthalate) to about 37% for the 0.9 wt% silica filled PEN composite. There is a

significant effect on the promotion of crystallinity of the polymer matrix according to the above-mentioned study [139]. However, in the present study on the PEEK polymer with 2.5-10 wt% silica or alumina, this effect was not seen. This might be partly related to the much higher level of particle content used in the current study. It is supposed that the higher the particle content in the polymer matrix, the less the mobility of the polymer chain segments would be. And the less mobility in polymer chain segments could, more or less, hinder the growth of the polymer spherulites. Another factor is the fact that the maximum crystallinity fraction of the PEEK polymer can only be 48% [178], and the crystallinity of the pure PEEK used in this study is already 39%. The increment by adding nanoparticles might be very limited.

As shown in Tables 3.7 to 3.9 and Fig. 3.45, the inclusion of nanoparticles could lower the required time for crystallization of PEEK segments. The time required for crystallization on PEEK segments at a cooling rate of 2.5 ^oC/min is 11.2 minutes; and it is only 5-8 minutes for the silica or alumina filled PEEK composites at the same cooling rate. It is obvious that the heterogeneous nucleation prevailed when the nanoparticles are introduced. Meanwhile, at high cooling rates (15-30 ^oC/min), the crystallization time will show no significant difference between pristine and the nanoparticle-filled PEEK segments. It seems the effect attributed by heterogeneous nucleation would be gradually diminished when the cooling rate is faster than the magnitude of the required crystallization time.

As shown in Table 4.8, a polymer filled with finer particles would result in smaller spacing between particles, and in turn the smaller spacing that would hinder the growth of the PEEK crystallites. It follows that a lower crystallinity of PEEK polymer is expected.

However, in comparing the Xc values for composites filled with the same amount of SiO2

nano particles but with different sizes of 15 and 30 nm, the finer particles would lead to a

slightly higher crystallinity, as shown in Tables 3.7 to 3.9. Also referring to Fig. 3.46, the PEEK polymer filled with finer filler, silica 15 nm, appears to require longer crystallization time to the PEEK segments. This effect seems to contradict the above argument.

Nevertheless, it should be born in mind the small the particle would also provide a larger specific surface area in which the PEEK molecules could deposit and crystallize, in the sense of heterogeneous nucleation. The nearly spherical surface of silica, as shown in TEM images, Fig. 2.2, could be the favorable sites for crystallization. As a result, the PEEK segments should have more time to crystallize, as expected.

As for the effect from the volume fraction, the 30 nm alumina composites would possess a lower volume fraction than the other two, resulting in larger spatial distances, as shown in Table 4.8. Moreover, the irregular surface of alumina, as shown in Fig. 2.2, would result in more sites for polymer molecules to deposit and crystallize. Consequently, the PEEK polymer filled with 30 nm alumina nanoparticles would bring about higher crystallinity, as shown in Fig. 3.47, as compared with that of the 30 nm silica.

4.8 Closing remarks

Through the extensive studies on the fabrications of the Mg/CF/PEEK laminated composites and the nanoparticulate filled PEEK composites, there are a number of new findings. Firstly, on the fabrication of the Mg/CF/PEEK laminated composites, it reveals that the fabrication by means of sandwiching the APC-2 prepregs and the Mg sheets could offer an effective and easy-to-process route to design the carbon fiber reinforced Mg laminates, as compared with that fabricated by the liquid metal infiltration Mg laminates. The successfully fabricated Mg/CF/PEEK laminated composites can reach near 100% ROM values for tension or bending. Moreover, the laminated composites can sustain their tensile strength up to

150^oC.

Secondly, on the fabrication of the nanoparticules filled PEEK composites, it is proved that the incorporation of the silica or alumina nanoparticles can improve the elastic modulus, UTS, and Hv values of the resulting PEEK nanocomposites by 20-50%, with the sacrifice of tensile elongation. Moreover, the dispersion of the nanoparticles in the PEEK matrix reveals reasonable dispersion, and there is no apparent interaction between the nanoparticle and the PEEK matrix. The PEEK nanocomposites with filler contents ranging from 2.5 to 10 wt%

show minor variation in Tm, which ranges between 332 and 340^oC. And the larger variation of Tc was found to range from 295 to 260^oC. It is also shown that the overall crystallization time, tc, and the crystallinity, Xc, would be decreased as the filler incorporated.