Thermal rounding for shape modification of high-performance polyetherketoneketone and reinforced polyetherketoneketone-carbon fiber composite particles

This study presents shape transformation of anisotropic high-performance thermoplastic polyetherketoneketone (PEKK) and carbon fiber reinforced powder composite particles (HT-23) by thermal rounding. The shape transformation is achieved by (partial) melting of the high-temperature thermoplast microparticles. Three different process setups are presented, investigating the impact of the source of heat supply on the resulting shape modification: using a directly heated sheath gas flow, an indirect heat supply through the reactor wall and a combined approach. Regardless of the chosen setup, a modification of the particle shape was observable. The most advantageous shape transformation was observed in the indirect heating approach. In addition, the enhanced shape transformation yields an improved free flow behaviour of the powders, as quantified by ring-shear experiments. Reductions of the unconfined yield strengths of the powders for high consolidation stresses as high as 18 percent for PEKK and 30 percent for the HT-23 are achieved. Thereby, processability of the powder in laser based powder bed fusion is enhanced, extending the range of available (composite) polymer materials.


Introduction
Powder flowability is a key factor in many industrial processes, e.g., in powder dosing during die filling or powder based additive manufacturing (AM), as especially for the latter, the resulting part properties are mainly depending on the bulk solid and particle characteristics of the provided feed material. In addition, the layer quality is strongly affected by the particle's shape [1-3]. For varying particle geometries (potato shaped, spherical, irregular) distinct parameters, e.g., packing density and layer quality, are found. Especially spherical particles are known to be advantageous for generation of defect-free layers [4]. Layers produced from irregularly shaped particles showed a lot of defects, due to less dense packing and worse flowability during powder spreading of these feedstocks: Beretta et al. [5] showed for PA12 and PEEK, that the influence of the layer deposition on the final part properties is severe. Parts manufactured using irregular shaped particles show larger porosities in the final component. Therefore, preferred materials to be applied in laser powder bed fusion (PBF) of polymers (PBF-LB/P) are characterized by smooth surface structures and (close-to-)spherical particle shape, which ensures proper flowability and packing density [6][7][8][9][10].
The chosen production route for AM powder materials strongly affects the resulting particle morphology. Comminution routes are known to yield angular, edged powders, whereas in melt emulsification and liquid-liquid phase separation (LLPS) close-to spherical shaped particles are achievable [2,[11][12][13].
The above-mentioned requirements regarding the particle shape narrows the choice of materials suitable for PBF-AM, as the feasibility of melt emulsification and LLPS is restricted by the identification of proper systems and especially poses several challenges for processing of high-temperature thermoplasts. To perform melt emulsification, specific thermal and rheological properties have to be met by the thermoplast-solvent system; for LLPS suitable moderate solvents need to be identified. Furthermore, for the mentioned liquid phase processes additional postprocessing steps are required, for example, the separation of the microparticles from the liquid phase, e.g., by filtration and subsequent drying.
There are not that many high-performance thermoplast feedstocks available, as the high mechanical strength of those materials makes it difficult to produce proper particle sizes for use in AM. The prevalent manufacturing route for these feedstocks is comminution, resulting in a large variety of particle shape [2]. Comminution is used for the generation of composite particles, as well. In consequence, for the applicability of high-performance thermoplasts in the additive manufacturing process post-processing steps are required to further enhance the powder properties. Further manufacturing routes for composite materials, e.g., carbonization, were proposed by dos Santos et al. [14]. Additionally, Puértolas et al. [15] investigated the impact of the mechanical blending of graphene nanoplatelet/UHMWPE composite particles on the resulting electrical properties. Hence, especially for composite particles, post-processing methods are inevitable for the subsequent adjustment of the particle shape with respect to the desired application.
For the improvement of the particle shape of AM powders, various post-processing routes have been established, for instance the post-processing of comminuted products via thermal rounding [4,[16][17][18], plasma functionalization [19][20][21] or dry coating [18,22]. Moreover, the spheronization of pellets by means of mechanical rounding using varying types of rotor-stator units is established [23][24][25][26][27], where the process can be performed in different geometries, horizontal and vertical. The materials investigated were microcrystalline cellulose (x 50,3 = 0.63-2 mm), stainless steel powder (x 50,3 = 43 µm), pharmaceutical materials (x 50 = 60 µm) and graphite particles (x 50 = 250 µm). Kondo et al. [28] investigated the influence of enhanced temperatures in mechanical spheronization of pharmaceutical material crystals, reporting superior shape modifications with increased temperatures. Pfister et al. [29] described a method to increase the bulk density of polymer powders and blends, where a vertical mixing unit is applied for mechanical treatment.
To the authors' best knowledge, application of the above presented spheronization approaches for the modification of high-performance thermoplasts is scarce. Until today, the investigations regarding the processability of PEKK and composites mainly focused on the melting and rheological behaviour of the powder materials. However, bulk solid properties of the feed material play an important role with respect to the layer deposition [7,[33][34][35] and, thus, the final mechanical properties of the sintered parts [3], as well. Within this study the rounding of high-performance thermoplasts, PEKK and carbon-fiber filled PEKK, in a heated downer reactor is investigated in the framework of a parameter study with respect to the achievable shape transformation and the resulting improvement in powder flowability.

Material specification
The polyetherketoneketone (PEKK) feed material (EOS Electrical Optical Systems GmbH, Germany) is a semicrystalline high-performance thermoplastic with a melting point of T Melt = 303 °C and a median particle size x 50,3 of 63.5 µm. PEKK is known to be inherently flame retardant and shows a low hygroscopicity [36,37]. The HT-23 material consists of PEKK with 23 wt. % encapsulated carbon fibers. The median particle size x 50,3 is 79.4 µm. The melting point is identical to that of PEKK. EOS GmbH and Advanced Laser Materials (ALM) Inc. provided the commercially available powders.

Determination of particle shape
For two-dimensional static image analysis of the particles, an optical light microscope (Axio Imager M1m (Zeiss)) was used. Particles were dispersed on a microscope slide and evaluated with the software Axio Vision (Release 4.6.3 04-2007) in transmitted light mode. To obtain the distributions for the form factors circularity (C) and roundness (R), the images were analysed using the Biovoxxel toolbox "advanced particle analyser" using the Fiji platform according to [38]. The amount of recorded particles is on the order of 15,000 particles in each measurement to provide quantitative results and to ensure statistical significance of the results.
The particle form factors were calculated as where A and P are the measured projected area and perimeter, respectively. The major axis of the bounding ellipse is denoted by x max . (1) Both shape factors are related to an ideal circle, with maximum circularity and roundness equalling 1. Circularity describes the ratio of the projected area to the perimeter, see Fig. 1. Thus, it is a measure for the particle surface roughness. The roundness, on the other hand, defines the aspect ratio considering a constant projected area.
For the additional analysis of the particle morphology, a scanning electron microscope (SEM) Gemini Ultra 55 (Carl Zeiss) with an accelerating voltage of 0.5 kV and a secondary electron detector for imaging was used. The sample images were recorded at appropriate magnifications.

Determination of the particle size distribution
Particle size analysis was performed by laser diffraction (Mastersizer 2000 (Malvern) using the Hydro 2000S wet dispersing module. For the measurement, a suspension consisting of the powder (0.2 g), distilled water and a small amount of sodium dodecyl sulfate (SDS, 98% (Merck)) was prepared, where the SDS provided the stabilization of the suspension. Stirring and ultrasonication ensured a homogenous dispersion of the particles and suspension stability.
One adverse effect, that occasionally occurs during thermal rounding, is agglomeration of the particles close to their molten state. To ensure the absence of those agglomerates, a comparison of the volumetric particle size distribution of the feed material and the samples processed via thermal rounding approach can be considered. Agglomerated particles yield in a strong enlargement of the former particle size distribution, as both, the fine and coarse fraction are nearly doubled. Thus, by observing a moderate enlargement of the obtained particle size distributions, agglomeration effects are scarce.

Determination of the flowability
The characterization of powder flow is performed using the ring shear cell tester RST-01.01 (Dr. Dietmar Schulze Schüttgutmesstechnik). The applied consolidation stresses were 1300, 2600 and 4600 Pa. The ratio between the consolidation stress 1 and the unconfined yield strength c is defined as the flow function ff c , according to Jenike [40,41].

Experimental setup
The thermal rounding of high-performance thermoplastic microparticles is conducted using a stainless-steel lab-scale downer reactor, presented in Fig. 2. As already mentioned above, two process setups have been considered in this work, whereas the provision of the required energy is different. For setup (a) the particles are directly heated in contrast to an indirect heating approach in setup (b). In the beginning of this section the reactor setup is described, followed by a detailed description of the differing heating approaches.
The 0.8 m-long reaction zone has an inner diameter of 0.1 m, whereas the quench zone has a larger inner diameter of 0.19 m at a shortened vertical length of 0.6 m, tailing in a cone. After the treatment, the coarse fraction is collected in the cone (setup (a)) or at the sintered plate (setup (b)), and the fine fraction in the laterally mounted cyclone.
A flat-ribbon feeder (SAG 420, Topas GmbH) enables the transport of the feed material, which is stored in a hopper. The attached screw conveyor continuously feeds the powder material into the particle injector, with a mass flow rate of 30 g h −1 . The particle injector nozzle comprises a doublewalled stainless-steel pipe with an inner diameter of 15 mm and an outer one of 26 mm. As the pre-heated sheath gas flow already encloses the particle injector, the double-walled concept prevents the particles from melting in an early stage, preventing adhesion of partially molten polymer to the particle injector wall.
Considering the heating of the particles for setup (a), a pre-heated inert sheath gas (nitrogen) enters the reactor at the top (gas heater GA00565, Horst GmbH). An additional heating element (HBS, Horst GmbH) is installed along the sheath gas pipes to prevent severe heat loss in between the gas heater and the reactor inlet. The temperature of the pre-heated sheath gas at the reactor inlet is controlled by a type-K thermocouple, installed above the particle inlet. The particles are enclosed by the pre-heated sheath gas and, thus, molten. To quench the molten particles, compressed air at room temperature is provided in the quench zone, with a fixed volume flow rate of 1.5 Nm 3 h −1 .
For setup (b) instead of the gas the heating pad (HMSG, Horst GmbH) provides the energy required for heating and (partial) melting of the particles. Therefore, Fig. 1 Schematic illustrations of the modification of a the circularity, and b the roundness (adapted from [39]) neither the gas heater nor the heating band are used for this setup. The energy generated by the heating pad is passed on to the particles via the reactor wall, indirectly melting the particles inside. The consolidation process is identical to that in setup (a).

Experimental plan
In the experiments, the influence of the process setup on the degree of shape modification is studied. As the optimum process temperature is supposed to be close to the melting point of the respective high-performance thermoplastic, the chosen process temperature was 320 °C. The presented process setups have been studied using three variations: (i) directly heating the particles via the pre-heated sheath gas, (ii) indirectly heating the particles via the heating pad and (iii) combining both processes. The respective temperatures of the heating elements have been adjusted accordingly, ensuring an identical process temperature of 320 °C. Hence, the directly heated approach was performed under a sheath gas volume flow rate of 7 Nm 3 h −1 , whereas the sheath gas volume flow rate of the combined approach was 3.5 Nm 3 h −1 , as the remaining heat energy required was provided by the heating pad. In all experiments, a total amount of 50 g polymer material was used.

Shape analysis via SEM and light microscopy
To identify the degree of shape transformation, SEM images of the untreated and treated powder samples are provided within this section. Figure 3 reveals a sponge-like structure for the PEKK raw material; the particles are irregularly shaped. The shape of the PEKK particles as provided shows them to most likely stemming from a pulverization from polymerization flakes. Considering the direct heating approach, a rounding of the arborescent structures can be observed, however, the surface structure is still rugged. For the direct heating approach, the residence time of the particles in the appropriate temperature range of the reaction zone was not sufficient to transform the surface structure. Moving on to the SEM images of the combined approach, the surface is partially molten and the rugged structure is developing a smoothened surface. By analysing the particles after the treatment with the indirect heating approach, for small particles (< 50 µm) ideal spherical structures with extraordinary smooth surface structures are obtained. Larger particles show a superior rounding of the former sponge-like pattern. In addition, the surface structure is partially smoothed. Considering the resulting shape modification, a very good spheronization was achieved. The milling process leads to irregular shaped and edgy HT-23 particles, yet the latter are already exposing a superior form as compared to the PEKK raw material. Via milling, angular fragments are created, which are characterized by a high surface roughness, see Fig. 4.
The thermal rounding via the direct heating approach mainly improves the surface roughness, whereas the impact on the brick-shaped structure is small. Considering the encapsulated carbon fibers, they are slightly less prominent, as the polymeric areas wrap around the fibers during the rounding process. For the combined process setup, an enhanced rounding of the brick-shaped structure is observable. Moreover, the formation of aggregated structures is less distinct compared to the PEKK powder samples. No distinct difference in between the direct heating and combination approaches is noticeable regarding the optimization of the surface structure. However, for the indirect heating approach the resulting particles show a close-to spherical shape. Furthermore, enclosing the carbon fibers is superior, as the surface tension of the polymer melt is inducing the particle to contract. Similar to PEKK particles, the most sufficient shape modification was achieved by the indirect heating approach, as both, energy transfer and residence time of the particles was improved. However, not all particles were completely affected by the spheronization process, as the energy supply through the reactor wall was not equally contributed over the cross-section. Therefore, particles remaining in the centre of the reactor may not be fully spheronized, because of the lower reaction temperatures. As this study serves as a proof of concept, further improvements of the reactor concept have to be considered for a thoroughly spheronization of the product.
The quantitative analysis of the particle shape is performed by means of light microscopy. The shape factors circularity and roundness of the two studied materials are given as number-based cumulative distributions in Fig. 5 for PEKK and Fig. 7 for HT-23.
As the feed material displays a sponge-like pattern with strong convexities, the resulting circularity is poor (avg. 0.58, max. 0.84). Throughout the thermal rounding process, the circularity increases up to C 50,0 = 0.72 and C 90,0 = 0.88 for the indirect heating approach. The weakest improvement of the particle circularity is observable for the direct heating approach. The process setup shows an insufficient particle residence time or the process temperature to achieve proficient melting of the particles. Regarding the combined approach, a proper increase of the shape factor circularity can be determined; however, on average the indirect heating approach is superior.
Regarding the roundness of the powder samples treated with the direct heating approach, similar observations can be made. However, the absolute values of optimization are lower, as the raw material already possesses proficient values considering the roundness. For both, the combined and indirect heating approach, the roundness distribution owns a special facet: in the beginning of the curves, the shape factor decreases, whereas starting from Q 0 = 0.5 the curves are showing a superior roundness. The decreased roundness can be attributed to the aggregates formed during the thermal rounding. As for those approaches, a major amount of the required energy is provided through the reactor wall, the areas near the reactor wall show a significantly higher temperature as compared to the centre regions. Thus, particles in the near-wall regions are in the molten state for an extended period of time, which in turn increases the chance of aggregate formation. However, those aggregates do not exceed the maximum particle size of the raw material, as the measurements depicted in Fig. 9 confirm. Additionally, the resulting shape modification is enhanced, as fifty percent of the investigated powder samples in Fig. 5 depict a superior roundness.
The dependency of shape transformation on initial particle sizes is depicted in Fig. 6. A higher density indicates a larger amount of the corresponding combination of shape factor and particle size.
For the PEKK raw material, proper shape factors for the circularity are predominantly observable for small particles with a diameter of 10 µm. However, the highest number of small particles is located in a range of C = 0.3-0.8. For larger particles, the shape factor circularity is decreasing, showing a range of C = 0.3-0.6. Regarding the roundness factor of small particles (below 10 µm), a broad distribution can be observed, whereas the major amount is located at a shape factor of R = 0.3. With an increasing particle size, the broadness is decreasing to a range of R = 0.4-0.9.
The direct heating approach is increasing the shape factors, especially for particles with a diameter with 20 µm and above. Moreover, the intensity of the particle amount with sizes around 20 to 40 µm is enhanced. The strongest modification of the density plots can be determined for the combination and indirect heating approach. For both process setups, the shape modification is performed across the entire particle size distribution, whereas the degree of shape modification is enhanced for the indirect heating setup. Furthermore, the presence of aggregated structures can be verified within those plots, as for the Fig. 6e-h, the number of particles at shape factors within the range of 0.2 to 0.4 increases. However, the maxima of the shape factors are more uniformly distributed throughout the particle size distribution.
The circularity is increasing during thermal rounding, yet, the indirect heating approach owns the major improvement, similar to the PEKK powder samples. However, the differences in between the remaining samples are moderate, as the HT-23 feed material already shows a proper circularity (C 50,0 = 0.67 and R 50,0 = 0.63).
Regarding the roundness of the particles for both, the direct heating and combined approach, a moderate decrease is observable. This decrease can either be attributed to the formation of minor aggregates, whereas for the direct heating approach the decrease is marginal, or to the more prominent exposure of the carbon fibers. The indirect heating approach depicts a superior improvement of the shape factor roundness. As confirmed by the SEM images, the carbon The HT-23 material already presents proper shape factor distributions for particles up to 20 µm, yet the width of the distribution is high. During thermal rounding a shift of the density distribution can be observed in Fig. 8. The initially broad distribution of both, circularity and roundness, is developing towards a more concentrated region, for the circularity at C = 0.6-0.9 and the roundness at R = 0.4-0.9, however, significant differences between the Fig. 8a-f are not observable. For the indirect heating approach (Fig. 8g and h) the density distribution of particles with diameters up to 45 µm is enlarged. In addition, the maxima of the shape factors are more concentrated at C = 0.8 and R = 0.6, especially for particle sizes around 15 to 20 µm. The density for particle sizes below 10 µm is reduced moderately.

Particle size distribution
The formation of aggregates during the thermal rounding process is further analysed by the investigation of the emerging particle size distribution of the untreated and the treated powder samples via the volumetric density distribution q 3 (x). The solid line in Fig. 9 represents the untreated a b Fig. 6 Distribution of the shape factors circularity and roundness with respect to the particle size of untreated PEKK (a, b), direct heating (c, d), combination (e, f) and indirect heating (g, h) raw material, whereas the dashed lines represent the treated powders.
Regarding the width of the q 3 -distr ibution ( span = (x 90,3 − x 10,3 )∕x 50,3 ), all samples show a similar span of around 1.3. Throughout the rounding process, the fine fraction is shifted towards larger particle sizes, whereas the number of fine fractioned particles remains constant. Furthermore, the minimal particle size is enlarged, from 3.8 to 5 µm. The authors assume, that the finest particles completely coalesced with slightly larger ones, as the fine fraction is almost identical for the three separate process setups. The decrease of the median fraction in combination with the increase of the maximum particle size, as again a shift towards larger sizes can be observed, leads to the assumption that a minor formation of aggregates was triggered.
With respect to the particle size the most successful rounding was performed within the combined approach, as on the one hand, the number of median fractioned particles is the highest and on the other hand, the maximum particle size observable is closest to the raw material. Thus, the enhanced residence time in combination with the indirect energy supply leads to a superior spheronization of largesized particles.
No differences in between the direct and indirect heating approach can be determined, as the thermal rounding induced the formation of aggregated structures, considering the lower number of median fractioned particles and major particle sizes.
Similar observations are made for the fiber reinforced HT-23 samples, presented in Fig. 10. The span decreases throughout the rounding process compared to the raw material. Comparable to the treated PEKK samples, the amount of the fine fraction decreases, whereas the middle fraction increases. However, as the particle size distribution is in general larger, no effects of discharged particles are noticeable. Yet, the particles with diameters around 23 µm decreases to a minimum amount. The authors assume that those particles coalesce with the fine fraction, shifting the middle faction to larger sizes. In general, for the HT-23 powder samples, coalescence effects are present, yet non-spherical aggregates have not been observed frequently in the SEM images. The maximum particle size remains constant.

Flowability of PEKK and HT-23 powders
Thermal rounding improves the powder flow behaviour as confirmed by ring shear cell measurements of the products (Fig. 11), additionally the exact values of the flow factors are provided in Table 1. However, at both, low and mean consolidation stress, the unconfined yield strength C of the powder sample treated with the direct heating process depict lower values in comparison to those of the PEKK raw material. This can be attributed to the enlarged maximum particle size in the treated powder samples along with the decreased amount in the median fraction. Yet, for the highest consolidation stress applied, the unconfined yield strengths C are superior. For the indirect heating approach, the unconfined yield strengths C for the lower consolidation stresses is aligning with the one for the PEKK raw material, thus, for the highest consolidation stress a distinct increase of flowability can be observed. For the combined approach, To further increase the flowability behaviour, post-processing via dry coating can be applied, as e.g., shown in [22] and [42] for coating PE-HD and PP with silica nanoparticles. Performing the dry coating process at enhanced temperatures further improves the resulting flowability behaviour of the feed material [42]. Another post-processing method is the in situ plasma enhanced chemical vapour deposition (PECVD), where microparticles are coated in a fluidized bed reactor using the plasma afterglow to produce silica nanoparticles from an organic precursor [20].

Process yield
The yield resulting from the thermal rounding is analysed within this section. The yield is defined as the powder mass remaining after sieving of the fine and coarse product is completed, thus, represents the usable amount of polymer powder for further analysis, as proposed in Eq. 3. The respective yields for the different setups and polymers are given in Table 2.
For the neat PEKK polymer, the highest yield was achieved for the combined approach, followed by the indirect heating setup, whereas the direct heating setup has the lowest yield, yet, the yield is consistently more than fifty percent. The HT-23 powder samples show an enhanced yield, with a maximum of 85% for the direct and a minimum of 75.2% for the indirect heating approach, as the combination has a yield in between. The percentage value within the minimum and the maximum is 13.8% for the neat PEKK and 9.8% for the HT-23 powders, showing a similar range. However, the carbon fiber reinforced composite polymer thoroughly owns a superior yield, as the formation of aggregated structures is scares, compared to the feed PEKK material.

Summary and outlook
In this work, high-performance PEKK and PEKK-carbon fiber composite (HT-23) powders were spheronized by thermal rounding in different setups. For the first time, the thermal rounding approach was successfully performed for a high-temperature thermoplast. For this purpose, three process setups were considered to investigate the impact of the energy supply on the degree of shperonization. SEMimages as well as the form factor analysis of the microparticles show an increase of the spheronization for each investigated process setup, however the indirect heating approach depicts a superior spheronization. Furthermore, the formation of aggregates is depending on the feedstock material, as for the PEKK powder samples the formation of aggregates is present, whereas for the carbon fiber reinforced HT-23 powder samples aggregates are extremely scarce. The free flow behaviour of the high-performance polymers was successfully enhanced, suggestion a superior processability in the PBF-LB/P approach. Therefore, regarding the concrete application of the different process setups, the following conclusion can be drawn from this study: for high-performance polymers, the direct heating approach yields in an insufficient functionalization, as the total residence time of the particles is crucial. The most Output mass of entire Q 3 (x) Input mass * 100 Fig. 9 Volumetric particle size density distribution of the untreated and treated PEKK samples. Mean particle size x 50,3 and span are given in the inset Funding Open Access funding enabled and organized by Projekt DEAL.
Data availability Research data will be made available upon reasonable request to the corresponding author.