Effect of tip ultrasonication on the PEEK particles
Acoustic cavitation produced during TS promotes highly intense local conditions characterised by short-lifetime hot spots whose temperature rises to 5000 °C, reaching pressure values of around 1000 atm, and cooling rates of 1010 K·s−1 . Even though propagation of ultrasonic waves in liquids is generally considered as an adiabatic process because of the rapid pressure variation , the highly intense local conditions can cause ceramic, metallic, carbonaceous, and polymeric particles to experience, according to their nature, changes of their molecular weight, chemical and crystalline structure, and morphology [32, 33, 61,62,63,64,65,66,67]. Hence, the first step was to determine the effects of TS on PEEK particles.
In principle, FTIR spectroscopy can be used to monitor variations in the molecular weight, and chemical and crystalline structure. FTIR spectra of neat PEEK particles is presented in Fig. 2a. The thermal degradation mechanism of PEEK involves random scission of ketone and ether linkages [68,69,70], leading to the disappearance of characteristic PEEK bands in mid-infrared (MIR) at 1652 cm−1ν(C=O), 1225 cm−1ν(φ–O–φ), 1010 cm−1ν(C–O–C) or ν(C–O), and 927 cm−1νsym(φ–(C=O)–φ), and, to the development of a band at 1709 cm−1 related to an aldehyde group, which is well defined even at weight loss values as low as 5% . In contrast to thermal degradation, the ultrasonic degradation of polymers in solution is coupled to a non-stochastic mechano-chemical shear; long polymer chains are, therefore, degraded much more rapidly than small ones, resulting in average molecular weight (Mw) reduction at earlier stages . In the present experiments, PEEK particles were insoluble in the ethanol/isopropanol medium, and the differences in the spectra for the neat PEEK and processed polymer after TS at constant amplitude values of 30 and 60% were modest (Fig. 2a). Moreover, there was no evidence of new bands associated with aldehyde or other groups as possible result of a thermal degradation mechanism; however, there was a slight intensity reduction and shift of PEEK characteristic bands after TS (see inset in Fig. 2a), with increasing amplitude. This behaviour can be attributed to Mw reduction of the PEEK particles, and, in last instance, to a small contribution from ultrasonic degradation of PEEK. Similar results have been found in ultrasonication of chitosan nanoparticles .
Crystallinity plays an important role in semicrystalline polymers, not only because it affects their final properties, but also for the EPD process itself. Amorphous polymer chains extending out of the particle surface make a major contribution to the suspension stability . The behaviour of insoluble polymer particles in suspensions is similar to that of solid particles such as metals and ceramics with adsorbed polymers on their surfaces [72,73,74]. TS is known to modify material crystallinity in some cases , but the effects on PEEK are not yet established. FTIR provides an alternative to estimate the degree of crystallinity; in case of PEEK, it is through the MIR bands at 1305 cm−1ν(C=O), 1280 cm−1ν(C=O), 966 cm−1ω(CH), and 952 cm−1ω(CH) [75,76,77,78,79]. By using these bands, Chalmers et al.  and Cebe et al.  established calibration curves to quantify the degree of crystallinity of PEEK from FTIR measurements. In this work, the calibration curve for the C-H wagging mode reported by Cebe et al.  is applied, as it uses the band area rather than intensity ratio and was cross-checked against a range of techniques to determine the degree of crystallinity. According to the FTIR measurements and the selected calibration curve, no significant variations in the degree of crystallinity of PEEK were detected under the TS experimental conditions (Table 3). The degree of crystallinity was also estimated from XRD measurements: Fig. 2b shows XRD patterns of the pristine PEEK powder before and after tip sonication considering amplitudes of 30 and 60%. PEEK has an orthorhombic unit cell where the polymer backbones are aligned with the c-axis, giving rise to four main peaks in the XRD diffractogram at 2θ values of approximately 19°, 21°, 23°, and 29°, assigned to the (110), (111), (200), and (211) crystalline planes, respectively [80,81,82,83]. The degree of crystallinity for pristine and tip sonicated PEEK powders was estimated from XRD measurements according to the procedure reported in the literature [81, 83]; the values are summarised in Table 3. A comparison between the degree of crystallinity measured by XRD and deduced from FTIR calibration curve  shows a good correlation between the two methods and suggests that the effect of TS on the crystallinity of PEEK is minor and can be neglected.
In addition to the particle’s structure, its overall morphology is a key factor in the processing and properties of composites and nanocomposites. During TS, polymer particles may change morphology, depending on both process parameters (power, amplitude, suspension concentration, suspension volume, and microtip diameter/length immersed) and the characteristics of the particles (initial size, and their chemical and physical characteristics). TS is commonly associated with reduction of particle size; however, under some conditions, TS increases the polymer particle size; for example, an increase in particle size has been observed in TS of poly(methyl methacrylate) particles, where the authors attributed the behaviour to the fusion of the beads in dumbbell-like structures . Light-scattering (LS) techniques have been used to estimate PEEK particle size after TS; nevertheless, agglomeration and precipitation of the particles cause fluctuations in the measurements [37, 39]. Furthermore, the effects of TS on PEEK particle shape have not been reported; thus, a detailed SEM study was performed here. A large number of SEM images (up to 10 images, more than 400 particles analysed in each image) was obtained (typical example in Fig. 3a) and analysed quantitatively to obtain a BPDD which offers valuable information about how the variables interact or are correlated with each other (Fig. 3b). The initial PEEK particles are defined by major and minor particle size values ≤ 40 μm. Tick marks situated close to the BPDD axes show the locations of the individual observations. The diagonal line indicates the location of equiaxed particles, with an aspect ratio of 1. Rounded particles (e.g. particle numbered 1 in Fig. 3a) appear near the line in Fig. 3b, and elongated particles (e.g. particle number 2 in Fig. 3a) lie further away. There are relatively fewer larger particles (major axis over ~ 20 μm), but those that are present tend to be more elongated (Fig. 3b).
Despite the fact that BPDD provides valuable information, noise in the measurement process and subinterval size can influence the result; for example, the probability that an observation lies in a subinterval depends on the subinterval size selected. By contrast, the use of KDE reduces these effects, allowing a fundamental data smoothing where inferences about the population are made . KDE values derived from the individual observations are shown in Fig. 3c. The representation of the experimental observations in terms of KDE shows a distinct broad peak, indicated by the arrow, whose coordinates are related to the estimated particle sizes with the highest probability. This peak is centred at 13 μm (major size) and 9 μm (minor size) values, implying an average aspect ratio around 1.4. An archetypal PEEK particle with the characteristics described above is indicated in Fig. 3a as number 3.
After TS, both BPDD and KDE show a significant reduction in modal particle size, as well as an increase in particles smaller than 20 μm (Fig. 3). KDE (Fig. 3i) highlights that particles below 5 μm are the dominant ones after the higher amplitude (60%) treatment. At 30% amplitude (Fig. 3f), the distribution is bimodal, and the larger peak shifts to lower size values in respect to the original particle distribution. The shifts of distribution were observed to be correlated with the amplitude (Fig. 3f, i); nevertheless, the particle size was reduced by only 30% though the amplitude was doubled. This behaviour is consistent with that observed in polymers and other materials, where the rate decreases with particle size as consequence of energy lost in particle motion, compression, flexing, and other mechanisms rather than fragmentation . Although the LS measurements in the case of PEEK particles are time dependent because of particle agglomeration and precipitation, the particle sizes determined by SEM are consistent with the range of values determined by means of LS of a few to tens of microns [37, 39]. Interestingly, TS generates more elongated PEEK particles, with the aspect ratio increasing from 1.4 for the original powder to 1.75 to 2.5 when the amplitude is increased from 30 to 60%, respectively. Table 4 lists additional statistical parameters for neat PEEK and tip-ultrasonicated PEEK particles.
After TS, the PEEK particle dimensions are closer to the GO flake size, 0.5–2.0 μm, and have a more elongated shape, which may be an advantage in their interaction with the 2D nanosheets. Although there were only minimal changes in PEEK molecular weight and crystallinity, extended TS can be undesirable. High amplitudes increase the unavoidable erosion of the microtip, introducing impurities and contaminating the suspension. Therefore, a 30% constant amplitude was selected for subsequent processing of the PEEK powder.
The suspensions of PEEK particles and GO were mixed and then deposited by EPD onto stainless steel substrates, to form high-quality, uniform coatings. SEM micrographs show that the GO nanosheets are homogeneously distributed across the coating (Fig. 4) and also suggest two characteristic modes of interaction between the two phases after EPD and subsequent drying process: nanosheets deposited on the top or trapped in between the PEEK particles (Fig. 4a, b); and nanosheets wrapping the particles (Fig. 4c). Most importantly, SEM micrographs indicate that GO nanosheets form a co-continuous structure (Fig. 4d–f). The large GO nanosheet structures have dimensions exceeding the flake size indicated by the manufacturer, 0.5–2.0 μm, suggesting an assembly process during the EPD. GO continuous morphologies  and GO alignment have been reported . Nevertheless, in the present conditions, though it is plausible that the position of the GO nanosheets could differ from the horizontal position, it is remarkable that the GO nanosheets basal plane is predominantly aligned with the nanocomposite coating surface, forming structures whose dimensions are larger than several thousands of μm. SEM at increasing magnification shows the assembly process in more detail (indicated by the arrows in Fig. 4d–f).
Alignment of GO nanosheets, as result of EPD, has been discussed in the literature [17,18,19,20]. One explanation is that the basal plane will align to the direction of the applied electric field during EPD due to the anisotropic polarizability of the GO nanosheets; another possibility is that the GO aligns parallel to the substrate either due to rotation during deposition or due to capillary effects during drying. Large and aligned discontinuous GO structures were formed with 0.5 wt% GO loading in suspension (Fig. 5a, d), but as the GO content increased to 1.0 wt%, the co-continuous morphology appeared (Fig. 5b, e), becoming more prominent at 3.0 wt% (Fig. 5c, f).
Regarding PEEK particle size and aspect ratio after EPD, it is not straightforward to quantify these parameters from SEM micrographs. Even by increasing the accelerating voltage from 1 to 10 kV (Fig. 6) with the aim of varying the transparency of the GO nanosheets to the electron beam [58, 87], allowing PEEK particles to be visualised underneath, GO flakes are seen to cover the particles, specially small particles. Nonetheless, the majority of the observable deposited PEEK particles have a size below 10 μm, exhibiting elongated shape of low aspect ratio (Fig. 5d–f), although there was a considerable population of particles (after TS) with major size ranging between 10 and 20 μm after TS (Fig. 3d–f). This effect is likely associated with the sedimentation of large PEEK particles during EPD. Composition, particle size and shape, and size distribution are key factors dictating the success of EPD [88, 89], confirming that TS is an effective strategy to prepare suitable suspensions containing PEEK powders and GO for EPD.
Macroscopically, PEEK/GO nanocomposite coatings deposited at 30 V were, in general, more uniform and thicker than those deposited at 10 V. Nevertheless, there was a significant edge effect when deposition time was extended over 4 min (Fig. 7), especially when the GO concentration increased in the suspension, leading to crack formation along the substrate edges. Therefore, samples prepared by EPD at 30 V and 3 min were selected for further analysis.
Electrophoretic deposition of PEEK requires a post-sintering process in order to densify the coating, as well as to enhance the adhesion to the stainless steel substrate [40,41,42, 44, 45, 49]. Digital images of the dried and thermally treated PEEK/GO coatings (Fig. 8) show coherent coatings. PEEK/0.5GO_TT and PEEK/1.0GO_TT were uniform in the centre of the sample, Fig. 8a, b, but a darker area was visible close to the substrate edges in both coatings. In contrast, PEEK/3.0GO_TT was relatively uniform. These macroscopic features may be related to the wettability and viscosity of the melted PEEK, being both phenomena, wettability and viscosity, dependent on the heat-treatment temperature . Moreover, GO can substantially increase the viscosity in polymers above the rheological percolation threshold , and, in PEEK/GO composites, the viscosity could be increased one order of magnitude with respect to neat PEEK even at GO concentrations lower than 5% . This suggests that the nanosheets impede the free spreading of the melted PEEK over the surface, specially, at the edges (Fig. 8), where it is likely that more GO is deposited due to field enhanced deposition (Fig. 7).
The uniform central region of the consolidated PEEK/GO_TT coatings (Fig. 8) was studied by SEM, XRD, and FTIR to explore the effects of GO. Increasing GO content creates a rougher, more irregular surface (Fig. 9a–c). These irregularities can be attributed to the GO nanosheet structure. PEEK particles wrapped by GO nanosheets after EPD (Fig. 4c) may not be able to integrate into the continuous film during melting. Comparison of the SEM micrographs of Fig. 5d–f with those in Fig. 9a–c shows that the size and distribution of the particles before thermal treatment correspond with the observed irregularities. Furthermore, the texture of the co-continuous GO phase may be also manifested in the coating surface (compare Fig. 5). In addition to the micron-scale texture, a finer dendritic morphology with preferential orientation appears on all samples in the polymer-rich regions, which decreases at higher values of GO wt%; meanwhile, the roughness increases (Fig. 9d–f). This dendritic morphology was also found in neat PEEK coatings after drying and thermal treatment (PEEK_TT) at the boundaries of the coating (Fig. 10).
Transcrystallinity or surface crystallisation is a condition where nucleation and crystalline growth are affected by an existing surface; as a result, spherulites originate at the surface and grow normal to it. Transcrystallinity has been reported for PEEK on different substrates [76, 80, 93], and more recently on GO . In PEEK/GO composites obtained by injection moulding, a PEEK dendritic morphology similar to that shown in Fig. 10 was observed growing with preferential orientation on a GO flake and it was related to changes in the ratio of the (110) to (200) peak intensities of the bulk XRD patterns . Analogously, the XRD pattern for the PEEK/3.0GO_TT coating (Fig. 11) showed a decrease in the ratio of the (110) to (200) peak intensities in comparison with those of pure PEEK and PEEK_TT, indicating an important increase in the crystalline phase as a result of the thermal treatment in both PEEK_TT and PEEK/GO_TT coatings. The degree of crystallisation for PEEK_TT and PEEK/3.0GO_TT coatings is summarised in Table 5. Thermal treatment tends to increase the degree of crystallisation of pure PEEK (Tables 3, 5) [77, 81, 83]; nevertheless, graphene appears to have a nucleating effect for the crystallisation of PEEK from the melt and to affect the crystalline growth as well [94, 95]. In the case of PEEK/3.0GO_TT coating, the degree of crystallinity appears to be reduced by the co-continuous GO structure within the PEEK composite (Table 5). A similar effect has been recently reported in a non-isothermal crystallisation study of PEEK/graphene nanoplate composites by synchrotron X-ray diffraction .
On the other hand, location and chemical structure of the surrounding polymer, in addition to time and temperature, have been observed to affect the in-situ reduction of GO in polymer matrix composites [96, 97]. In this respect, the FTIR spectrum of neat GO after thermal treatment, GO_TT, is shown in Fig. 12, and it is compared with those of PEEK_TT and PEEK/3.0GO_TT. In the inset of Fig. 12, the contribution of GO_TT and PEEK_TT to the PEEK/3.0GO_TT spectrum is presented. During thermal reduction of GO, the restoration of the aromatic rings gives rise to the band at 1570 cm−1, indicated in the inset of Fig. 12, which is associated with C = C stretching vibration of the sp2 hybridised carbon atoms . Additionally, the thermal treatment also induced changes in those bands associated with crystallinity [75,76,77,78,79]. However, there was no sign of the aldehyde band at 1709 cm−1 which is associated with the onset of thermal degradation of PEEK , thus confirming an adequate thermal treatment adopted in this work. The degree of crystallisation of PEEK_TT estimated by considering the calibration curve  agreed with the XRD measurements (Table 5).
Another important characteristic is that the morphology of the PEEK/3.0GO_EPD coating developed at 10 V and 1-min deposition time (Fig. 4f) and that of the coating obtained at 30 V and 3 min deposition time were similar (Fig. 5f). Thereby, it is reasonable to conceive the characteristics of the coating morphology across the cross section as being similar to those on the coating surfaces. This conception of the coating morphology is supported by SEM images of the coating cross sections (Fig. 13), for PEEK/3.0GO_TT prepared at deposition time of 3 min, highlighting the consistent microstructure throughout the EPD film. The texture observed on the upper surface of the films is due to the presence of the wrapped PEEK particles (Fig. 13a, b). In addition, similar to the surface morphology (Fig. 9a–f), both wrapped PEEK particles and PEEK dendrites are observed (Fig. 13c, d).