2.1 Materials

The HMWPE used in this study was GUR 4118 (Celanese, Frankfurt am Main, Germany). The virgin GUR 4118 has a viscosity average molecular weight M_v of 6 × 10⁵ g mol⁻¹ [38]. Moreover, the carbon black used in this study was KLAREX RG 225 (Himadri, India), which was supplied by Weber & Schaer (Hamburg, Germany).

The GUR 4118 was dry-blended with different weight concentrations of CB by putting both materials in a bucket and subsequently shaking for 5 min. The investigated powder mixtures are shown in Table 1. The pristine GUR 4118 is used as reference (GUR 4118-Virgin).

2.2 Powder Properties

The combination of optical microscopy and the image software ImageJ is used to determine the particle morphology, the circularity, and the particle size distribution, respectively. The used light stereo microscope was manufactured by OLYMPUS type DSX 500. The lens used has a 50x magnification. Particles of each powder mixture were placed on the object desk. To record at least 100 particles for each powder mixture, microscopic images were recorded at different positions. The software ImageJ converted the recorded images to a black white image. Care was taken that the particles were separated from each other in order to measure the circularity and particle size for each particle.

The bulk density of the powder mixtures is determined after DIN EN ISO 60 [39]. 110 mL of the powder mixture was filled in the funnel and then flowed into a cup with a volume of 100 mL. Afterwards, the mass of the powder in the cup was measured to determine the bulk density. The measurement is done three times for each powder mixture.

2.3 Fourier Transform Infrared-Spectroscopy

The investigation of the optical properties and Fourier transform infrared (FT-IR) spectra in the attenuated total reflection (ATR) mode of compression molded parts with different carbon black concentrations was determined using an FT-IR spectrometer Frontier (Perkin Elmer, Massachusetts, USA). For the measurements, 64 scans were collected and averaged with a resolution of 4 cm⁻¹ in a wavenumber range from 4000 to 650 cm⁻¹. All measurements were carried out at room temperature.

2.4 Differential Scanning Calorimetry

A differential scanning calorimetry (DSC) apparatus (type DSC 8000 of Perkin Elmer, Massachusetts, USA) was used to determine the sintering window of the different powder mixtures. Therefore, one heating and one cooling cycle were performed. The measurements were performed in a nitrogen atmosphere, and the mass of the powder was approximately 5 mg. A temperature range from 30 to 190°C was chosen for the investigations. At the start and the end temperatures a holding time of 5 min was chosen to reach thermal equilibrium. The used heating and cooling rates were both 10 K min⁻¹.

Since the carbon black might change the thermal behavior during the process and affect the temperature distribution, the specific heat capacity c_p of the powder mixtures was determined in accordance to DIN EN ISO 11357-4 [40]. The investigated temperature range was also from 30°C to 190°C with a heating rate of 10 K min⁻¹.

2.5 Rheological Experiments

The rheological experiments were carried out in shear using a rotational rheometer (Kinexus Ultra+, Netzsch, Selb, Germany). Cylindrical samples with a diameter of 8 mm and a thickness of 1 mm were prepared by compression molding at a temperature of 200°C and a compression time of 6 min. The frequency sweeps were carried out at a temperature of 150°C. After insertion of the samples, a time interval of 5 min was chosen for thermal equilibration. The gap was set by using an initial normal force of 0.5 N for 1 min to realize complete contact between the plates. To determine the linear viscoelastic regime, stress sweeps at an angular frequency ω of 10 rad s⁻¹ were carried out between 100 and 10 000 Pa. Frequency sweeps were subsequently performed in the range of angular frequencies from 0.01 to 100 rad s⁻¹ (starting with the highest frequency) with a stress amplitude of 1000 Pa. All tests were carried out in a nitrogen atmosphere.

2.6 Mechanical Properties and Coalescence of Laser-Sintered Parts

Tensile tests under uniaxial loading with tensile bars of type 1BA were carried out to determine the mechanical properties. For the tensile tests, a test rig from Shimadzu type EZ-SX with a force transducer of 5 kN was used. The crosshead speed was 1.2 mm min⁻¹, which is equal to a strain rate of approximately 2 % min⁻¹. The nominal strain rate was determined by taking the movement of the crosshead into account. The nominal stress σ was calculated by dividing the measured force by the initial cross-sectional area of the sample. All tensile tests were done under ambient conditions (room temperature). The Young's modulus E is determined after DIN EN ISO 527-1 in the strain interval from 0.05% to 0.25% [41].

Hereby, L_c describes the contact length between two neighboring particles. The variables P₁ and P₂ are the perimeters of the particles. For the investigations, 20 particle collectives were tested.

2.7 PBF-LB/P System

Processing of the powder mixtures was carried out on a PBF-LB/P system of the type Lisa (Sinterit, Krakau, Poland). The Lisa apparatus has a print bed size of 150 × 200 × 150 [mm] and the build chamber can be heated up to 210°C. Moreover, the piston, feed bed, and print bed can be heated up to reduce thermal gradients. With this device, it is possible to use layer thicknesses between 0.75 and 200 µm. For the powder deposition during the process, Lisa uses a counter-rotating roller recoater. In terms of the laser system, Lisa uses a diode laser with a power of 5 W and a wavelength of 808 nm [43]. Hence, it is necessary to add CB to GUR 4118 since polymers are only able to absorb the wavelength of 808 nm to a large extent with additivation. One of the typical additives to improve the absorption of polymer powders is carbon black [15].

The general processing of the powder mixtures was first tested using GUR 4118-CB(0.25). The laser power of the machine was fixed at 5 W. Therefore, changes in the volume energy density were achieved by adjusting the scan speed. For the initial laser sintering trials, a constant layer thickness of 125 µm and a hatch spacing of 0.3 mm were used, with a volume energy density of 260 J cm⁻³. This corresponds to a scan speed of 512.8 mm s⁻¹. The main objective of these initial trials was to identify suitable processing temperatures, which would then serve as a basis for further optimization of the processing parameters. A temperature range of 130°C–135°C (in 1°C increments) was investigated using job layout (a). After determining the appropriate processing temperatures, job layout (b) was tested using the same parameters as layout (a), with the aim of evaluating whether the reduced scanning area could help minimize curling (both job layouts are depicted in Figure 9). Additionally, qualitative trials were conducted with layout (b), using a layer thickness of 200 µm and hatch spacings of 0.2 mm and 0.3 mm, at the same constant energy density of 260 J cm⁻³. These tests aimed to assess the influence of layer thickness and hatch spacing on curling behaviour and to prevent curling during further processing. The final laser sintering trials were conducted at a print bed temperature of 135°C and a chamber temperature of 134°C, with a hatch spacing of 0.2 mm and a layer thickness of 200 µm. Four volume energy densities, 260, 340, 420, and 500 J cm⁻³ were tested, corresponding to scan speeds of 480.8, 367.7, 297.6, and 250.0 mm s⁻¹, respectively.

3.1 Particle Morphology and Size Distribution

The particle size distribution and the particle morphology play a key role for a successful PBF-LB/P process. The typical particle size of commercial PBF-LB/P materials ranges between 20 and 80 µm [44, 45]. Hereby, the content of smaller particles should not be too high since it could cause some agglomeration of the particles and hence lead to problems during recoating [15, 45, 46]. Regarding the circularity of the particles, this should be as high as possible since it leads to a higher powder density [15, 46-48]. The CB particles adhere to the surface of the GUR 4118 powder particles. Hereby, the distribution of the CB particles is inhomogeneous (Figure 2), which could lead to non-uniform melting of the particles, since a higher CB concentration yields a more pronounced absorption of laser energy.

The cumulative particle size distribution increases with increasing carbon black concentration. The carbon black particles on the surface of the GUR particles shift the particle size distribution to higher values (Figure 3a). In general, the particle size of each powder mixture is larger than the typical particle size for PBF-LB/P. For each powder mixture, the d₉₀ value is larger than 100 µm. Consequently, a layer thickness larger than 100 µm is necessary. Typically, laser sintering machines can accommodate a range of layer thicknesses, e.g., 75 to 200 µm, so powders with particle sizes larger than 100 µm are generally suitable. In addition to that, the circularity of the particles remains almost constant, whereas the bulk density increases up to a defined CB concentration (Figure 3b) and decreases for higher CB concentrations. The bulk density shows a significant increase by adding 0.25 wt.% CB. Hereby the smaller CB particles could fill possible voids between larger particles. This leads to denser powder layers during the PBF-LB/P. By increasing the CB concentration up to 1.00 wt.% the bulk density decreases, but is still higher than for the virgin powder. With further increase of the CB concentration, the bulk density decreases to a value that is smaller than for the virgin powder. In that case, the CB yields a higher tendency for agglomeration, which results in more voids. The bulk density is a function of the density of the single particles (i.e. approximately the density of HMWPE) and the density of CB, as well as porosity. Since porosity is a complex function of the sizes and the flow behaviour of HMWPE particles and CB, as well as the CB distribution, the data appear reasonable.

3.2 Optical Properties

The addition of carbon black increases the absorption of the powder mixtures for processing in PBF-LB/P (Figure 4). The FT-IR spectrum shows an increase of absorption at 943 cm⁻¹ with increasing carbon black concentration. The corresponding wavenumber of a diode laser is at 12376 cm⁻¹. This wavenumber is not covered by the FT-IR device. However, the tendency shows that the absorption is increased for each wavenumber except the ones where the characteristic polyethylene peaks appear (e.g., at 1454 cm⁻¹) [49-51]. According to the literature, CB shows a high absorption in the relevant wavelength range from 800 to 950 nm and increases with increasing CB concentration [52]. Therefore, it can be assumed that the absorption is also increased in the wavenumber range of the diode laser.

3.3 Sintering Window and Specific Heat Capacity

One key parameter for PBF-LB/P is the so-called sintering window. The sintering window describes the temperature difference between the onset temperature for melting and crystallization [15, 45, 53]. The addition of carbon black only has a marginal impact on the sintering window because it has no impact on the molecular level. The sintering window for all powder mixtures is approximately 10°C and hence very narrow (Figure 5). For comparison, polyamide 12 (PA 12) has a sintering window in the order of 25°C to 30°C [45, 54]. Moreover, the crystallization temperature strongly depends on the cooling rate. For lower cooling rates, the crystallization temperature is shifted to higher temperatures [15]. During the PBF-LB/P no constant heating or cooling rates appear in comparison to DSC measurements. Therefore, the sintering window during PBF-LB/P presumably is even smaller. To achieve a high coalescence of polymer particles and a high adhesion between the sintered layers, crystallization should be inhibited as long as possible [15, 45, 53]. If crystallization starts to early, then curling and warpage occur, which could lead to a potential collision between sintered layers and the recoater. Hence, selection of the optimum temperature and realization of a mostly spatially uniform temperature distribution are essential for successful processing of the powder mixtures.

3.4 Rheological Properties

The low CB concentrations do not affect the particle coalescence. Generally, the storage and the loss moduli G′ and G′′ of the pure HMWPE and the composites decrease with decreasing angular frequency ω and attain similar values, such that the loss tangent tan δ does not vary much with ω. A higher CB concentration only has a minor effect on the rheological properties for the investigated concentrations and attains similar values for each CB concentration within statistical scatter (Figure 7a,b). The average relaxation time τ can be roughly estimated by the crossover-frequency of 1.585 rad s⁻¹ for each concentration, which corresponds to an average relaxation time of approx. 0.63 s. The terminal zone of the Maxwell model is not attained in the investigated range of frequencies. On the contrary, the complex shear viscosity decreases with ω with the power-law |η*|∝ω^−0.5, independent of the CB concentration (Figure 7c). According to the literature, the rheological properties of polymers start to be affected by the CB concentration only at much higher concentrations of around 5 to 10 vol.%, respectively [55, 56].

This model correlates the half neck radius y and the particle radius r with the surface tension Γ, the time t, and the zero shear rate viscosity η₀. Therefore, the complex shear viscosity at low frequencies is one of the key properties for a fast coalescence. Hereby, a higher magnitude of complex shear viscosity leads to a slower coalescence and, thus in general, a weaker bonding between particles and laser-sintered layers during the manufacturing process. The zero shear rate viscosity η₀ cannot be determined in the investigated frequency range for the polymer/powder mixtures (Figure 7c). As the viscosity increases with decreasing frequency, it can be assumed that the zero shear rate viscosity of each powder mixture is higher than the experimental data at 0.01 rad s⁻¹. For comparison, PA 12 has a zero shear rate viscosity in the order of 1000 Pa s (depending on molecular weight and temperature) [57] and hence at least three decades lower than the zero shear rate viscosity of HMWPE. This is one of the major challenges, besides its narrow sintering window, of HMWPE for processing by means of PBF-LB/P.

3.5 PBF-LB/P Processing

An increased powder bed temperature inhibits the re-crystallization for a longer time. Therefore, the tendency to curl is slightly reduced (Figure 8), cf. the work of Mousa et al. [58] In the temperature range between 130°C and 133°C no sufficient bonding between sintered layers is achieved. The tendency to curling is too high and leads to displacements of the sintered layers, because of collisions with the recoater. A further increase of print bed temperature, above 135°C, was not tested. The reason was that 135°C is already outside the sintering window of the powder mixture and near the melting peak. A further increase contains the danger that a high amount of powder could melt. However, even at a print bed temperature of 135°C, some curling and insufficient fusion of the powder remain, which also leads to an insufficient bonding between adjacent layers (Figure 8f). This indicates that the cooling rates during the process are low enough that even at a temperature of 135°C re-crystallization could occur and that the energy input is not high enough to create a sufficient interlayer bonding. Especially previously printed layers will have a lower temperature than the top layer of the print bed, which leads to the fact that re-crystallization starts even faster, and the manufactured layers tend to curl and delaminate. To overcome the problems of curling and insufficient bonding between layers, the scanning area is reduced by using job layout (b) (Figure 9b). Since the initial trials showed that even at 135°C, curling occurs for job layout (a), job layout (b) was just tested at this temperature.

The reduced scanning area leads to a stronger bonding between the layers, since there are no voids between them (Figure S1). The energy input is high enough to form a sufficient interlayer bonding, since the energy input per area is increased. Nevertheless, curling still occurs. To reduce the curling further, the layer thickness was changed to 200 µm and the hatch spacing to 0.2 mm. The decreased hatch spacing causes a higher energy input, which in general should lead again to a higher temperature and inhibit the re-crystallization for a longer time. The increased layer thickness reduces the collision potential between the sintered layers and the recoater. Even if a certain degree of curling is present, the probability of sintered layers protruding from the powder bed is reduced due to the higher layer thickness. The laser sintering of specimens with a layer thickness of 200 µm and a hatch spacing of 0.2 mm was in general, successful (Figure 10). The CB concentration significantly effects the laser sintering result. An increase of CB concentration leads to a reduced processability (Figure 10). Especially at CB concentrations of 1.0 and 2.5 wt.% the tensile bars show deviations from the target geometry. For these tensile bars, it is not possible to perform tensile tests because of the geometrical deviation. Even for GUR 4118-CB(0.5) it can be observed that one tensile bar shows defects. It appears that pre-sintered layers are somehow colliding with the recoater and becoming displaced within the powder bed. However, no recoating defects, such as stripes in the powder bed caused by the movement of agglomerates, were observed. Only the processing using a CB concentration of 0.25 wt.% leads to tensile bars in sufficient quality. Even if a small degree of curling remains, the bonding between the layers is high enough that no empty voids between them can be observed. The reason for the observed defects could be the lower specific heat capacity for higher CB concentrations. This could lead to a faster cooling/discharging of the heat of laser-sintered layers and lead to an earlier re-crystallization, which results in curling. Those curled layers would collide with the recoater and be ripped out of the powder bed, which could lead to the observed defects at the tensile bars (Figure 10b–d). Another possible reason could be an increased expansion of the material in the build direction. Investigations of UHMWPE with different hatch spacings showed that this expansion becomes stronger when the hatch spacing decreases and is directly connected to the peak temperature [27]. A smaller hatch spacing corresponds to a higher energy input/temperature, because the overlapping of laser traces becomes larger. A higher CB concentration also leads to a higher peak temperature, because of the higher laser absorption [37]. This could also lead to an increased expansion in build direction, shortly after laser exposure, and therefore an increased potential of collisions between the recoater and sintered layers.

The used values for the densities are 950 kg m⁻³ for ρ_HMWPE and 2000 kg m⁻³ for ρ_CB according to the supplier.

Hereby, the density of the powder mixtures is also estimated by using the rule of mixture. Equation (7) implies that the thermal diffusivity increases by approximately 10% from 0.25 to 2.5 wt.% CB (Figure 11). Even if this increase in the time scale of temperature decay does not seem high, in the case of the investigated HMWPE powder, every temperature deviation could have a significant impact on the processing result, because of its narrow sintering window. This simplified approach shows that selecting an adequate CB concentration is important, because on the one hand, the absorption increases (and hence melting proceeds more rapidly), and on the other hand, a too high concentration affects the processing result.

3.6 Mechanical Properties and Coalescence

A higher volume energy density E_v leads to a stronger bonding between powder particles and adjacent layers. Thus the Young's modulus, nominal stress, and strain at break increase with increasing volume energy density (Figures 12 and 13). Hereby, a nearly linear dependency between the mechanical quantities and the volume energy density is observed. Experimental and numerical investigations have shown an almost linear relationship between energy input and the dimensions of the laser trace or weld. As the energy input increases, the laser trace becomes wider [37]. In the case of PBF-LB/P, this leads to an increased overlap of scan lines at a constant hatch spacing. As a result, more powder is melted, which contributes to improved mechanical properties. All tensile bars show brittle fracture behavior and therefore, the tensile strength corresponds to the stress at break. The brittle behavior of the specimens results from the layerwise manufacturing process and the remaining porosity. Moreover, the mechanical properties are affected by the part position in the print bed and the distribution of the CB particles. Both affect the energy input. The influence of the part position can be seen by the values of the standard deviation (Figure 12) and the exact nominal stress-nominal strain curve for each tensile bar in the (Figure S2). The position dependency of the mechanical properties results from two reasons. The first reason is that the temperature distribution is not spatially uniform, which results in different temperatures on top of the powder bed. Consequently, melting is more complete in areas where the powder bed temperature is higher. The second reason is that the energy input also depends on the shape of the laser spot. Typically, the spherical shape of the laser spot is reached exactly when the laser spot meets the center of the print bed. The deflection of the laser spot due to the mirror movement changes its shape. This leads to a more ellipsoidal shape when the laser spot is moved away from the center. This ellipsoidal shape has a higher area than the spherical shape and therefore, the energy density is lower for parts that are not in the center of the print bed. The absorption properties of the powder have a higher impact on the mechanical properties than the pure energy input. The different absorption properties (Figure 4) lead to differences in the mechanical properties, because the melting process is influenced by the absorbed laser energy. A higher CB concentration increases laser absorption and, consequently, raises the melt temperature. According to the literature, the peak temperature also shifts toward the interface of the absorbing material [37]. This promotes stronger bonding between adjacent layers and particles, as the decreased viscosity at elevated temperatures favors coalescence. With constant volume energy density, the mechanical properties increase with increasing CB concentration (Figure 12). To achieve mechanical properties similar to 0.50 wt.% CB the volume energy density for 0.25 wt.% CB needs to be increased up to 420 J cm⁻³.

A strong fusion of particles is one of the key factors to achieve high mechanical properties. In general, complete coalescence is not achieved during PBF-LB/P but the formation of so-called sintering necks [15, 42, 45, 62]. The formation of the sintering necks influences the density of the manufactured parts and consequently has an impact on the mechanical properties [15, 45]. Since every tensile bar shows a brittle behavior, it can be assumed that the interphase between neighboring particles is weak because of pores and only partial interdiffusion of polymer chains. The volume energy influences the degree of coalescence as well as the remaining porosity. Hereby, the degree of coalescence slightly increases with increasing volume energy density (Figure 13). The optical micrographs (Figure 14) reveal that no complete coalescence is achieved. With decreasing volume energy density, the general morphology of the powder particles remains. In addition, the degree of porosity is visualized by the voids (red circles) in the optical micrographs. Moreover, with the assumption that the volume of the tensile bars only slightly changes, the mass can be used as an indicator for the density. The mass increases with increasing volume energy density (Figure 15), and the porosity of the cross-sectional area decreases with increasing volume energy density. The low degree of coalescence and high porosity result from the dominant elastic properties/high melt viscosity of the polymer itself.