Investigation of porosity behavior in SLS polyamide-12 samples using ex-situ X-ray computed tomography

The system employed for the manufacturing of the test samples using SLS technology was EOS Formiga P110 (EOS GmbH, Krailling, Germany), and the feedstock material was polyamide-12 powder (PA2200) supplied by EOS. Refreshed powder was used, which is a mixture in a proportion of 50:50 (used: virgin). The manufacturing process was carried out using dedicated parameters (provided by the manufacturer) for this type of powder, with the processing temperatures being set to 166.5°C on the print surface, and 148°C in the removing chamber. The samples (five for each series) were produced (according to ISO-527 type 5b ‘Plastics – Determination of tensile properties – Part 1 and 2’) in two orientations in relation to the build direction – horizontal (XY) and vertical (ZX) (Figure 1A). Pictures of the samples after the ex-situ tensile test are shown in Figure 1B.

Fig. 1

To obtain the 3D geometry of the polyamide-12 samples, the XCT dual-tube system phoenix v|tome|x m (GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany) was used. The system is equipped with a 300 kV unipolar microfocus X-ray tube, a 180 kV transmission target nanofocus X-ray, and a high-contrast digital flat panel detector (GE DXR250). During the tests, the nanofocus X-ray tube was used, in which the molybdenum-tungsten target optimized the generated X-ray spectrum at low acceleration voltages and enhanced the detectability of low density scanned materials. The scanning parameters were: a voltage of 100 kV, a current of 320 μA, a magnification of 19.9, and a voxel size of 10 μm. A standard scan was made with a 360° rotation, with the number of projections equal to 3,000, each being the average of three single projections taken with an exposure time of 100 ms. Data reconstruction was carried out using dedicated software (phoenix datos|x 2.7.2) with default reconstruction parameters. The grayscale images were analyzed in VGSTUDIO MAX 3.3 (Volume Graphics GmbH, Heidelberg, Germany). A local segmentation method was used to obtain volumetric models of the scanned samples.

The tensile tests were performed on the Mecmesin Multitest-I (Mecmesin Ltd, West Sussex, United Kingdom) test frame using a 1 kN load-cell (Figure 2B). The samples were mounted in wedge tensile grips, which provide better clamping with an increasing force. The tests were carried out with a 10 mm/min speed, and were continued until the samples’ destruction. The test frame was connected to Emperor™ Force (Mecmesin Ltd, West Sussex, United Kingdom) software, which allowed for stress–strain characteristics to be created. The X-ray 3D reconstruction was carried out for each sample before and after the mechanical tests (Figure 2A, 2C). The samples were scanned in a stack of five at a time, but it was only the measurement area of the sample, together with a fragment of the grip part, that was subjected to XCT tests. The volumetric models of the samples obtained from the X-ray 3D reconstruction (before the tensile test) were separated and placed in the same coordinate system and at the same starting point, which is also the geometric center of the sample (Figure 2D).

Fig. 2

The volumetric models obtained for the samples after the performed static tensile test were initially manually fitted to the geometry of the sample, which was scanned in its original state. The ‘best fit’ option was then used to obtain a perfect match to the initial geometry. In the next step, the top part of the sample was translated (in the given coordinate system in the y direction) by the value of absolute elongation. This approach makes it possible to compare the results before and after the tensile test in a reproducible way.

Fractographic analysis using the microscopic method was carried out on fractures of the samples after the static tensile test. Analysis was performed on transverse planes of vertically and horizontally oriented samples using the Sigma VP 600 Scanning Electron Microscope (SEM) equipped with a backscattered electron (BSE) detector (Carl Zeiss, Oberkochen, Germany). For better imaging resolution and to minimize the charging effect, the samples were placed on a conductive carbon strip attached to a grounded sample holder and then covered with a thin layer of gold using a vacuum sputter coater (Quorum Q150R ES, Lewes, UK).

The stress–strain characteristics obtained during the static tensile test are shown in Figure 3. It was not possible to determine the yield point for the produced samples, owing to the fact that the transition between the elastic and plastic behavior of the material was smooth. After reaching the stress value at the level of about 50 MPa (Figure 3A) for the horizontally manufactured samples, a necking could be observed, which consequently led to their complete failure. For the vertical direction, the samples behaved differently (Figure 3B) – the fracture was more brittle, and necking was not present. The lack of the neck additionally translates into the strain value, which for the ZX samples was approximately two times lower when compared to the XY samples. This contributes to lower mechanical properties than those produced in the XY direction.

The obtained results are similar to the tensile strength values presented in the literature [20, 21].

Fig. 3

Ex-situ XCT characterization enables changes in the structure of the samples, both before and after the static tensile test, to be compared. The result of the porosity analysis, for one representative sample from each build orientation, and with the pore diameters marked in the form of colors, is shown in Figure 4. For the XY sample, porosity was 2.67% ± 0.14% and after the static tensile test it increased to 3.14% ± 0.16%, while for the ZX sample, porosity was 3.40% ± 0.08% and after the static tensile test it increased to 3.45% ± 0.09%.

Fig. 4

Figure 5 shows, in a bubble chart, the change in porosity before and after the fracture. Each defect is shown as a circle projected onto the XY plane, and the size refers to their diameters. For the samples built in the XY direction (Figure 5A), a homogeneous dispersion without the anisotropic distribution of defects was observed. There was a visible increase in the diameter of the pores after the tensile test. For the samples built in the ZX direction (Figure 5B), the dispersion of the defect was also observed in the samples; however, the characteristic growth of pores after the tensile test and their necking at the fracture area could not be observed. The anisotropy of the pore distribution related to the manufacturing direction is characteristic, i.e., the pores are arranged in layers, which indicates that the places of interlayer connections are the main cause of their occurrence.

Fig. 5

Figure 6 shows the quantitative change of porosity corresponding to the height of the sample's gauge sections. The blue line shows changes in porosity before the static tensile test, and the red line after the fracturing of the samples. The porosity of the XY samples in the initial state ranged from 2% to 3.5%, and was homogeneous for the measured length (Figure 6C). After the static tensile test, the porosity was greater at a height of −4 mm to 6 mm of the sample, reaching a value of over 8% in the fracture area. In the case of the sample from the ZX series, the porosity for the subsequent layers ranged from 2.5 to slightly over 5% before and after the tensile test, and no significant changes in the porosity level were observed around the fracture area (red line in Figure 6F).

Fig. 6

The number of newly registered pores after the static tensile test increased by 14% for the XY samples, and by 0.8% for the ZX samples. Most of the newly recorded pores that appeared after the tensile test were <50 μm in diameter (Figure 7). In the range of diameters between 60 μm and 190 μm for the samples built in the XY orientation (Figure 7A), and 60–80 μm for the samples built in the XZ orientation (Figure 7B), a smaller number of pores were observed for the samples after the static tensile tests. In this range, the pores coalesce to the greatest extent and there is a perceptible increase in their diameter.

Fig. 7

The change in the shape of the pores can be described by a parameter that determines their sphericity, which is expressed by the following relationship (Eq. (1)): (1) Sphericity=ASphereAPore \text{Sphericity}=\frac{{{A}_{\text{Sphere}}}}{{{A}_{\text{Pore}}}} The closer the value of sphericity is to 1, the more the shape of the analyzed pore corresponds to an ideal sphere. The change in this value for all the registered pores in the tested samples is shown in Figure 8. For the XY orientations, an increase in sphericity was observed (Figure 8A). For the ZX samples, a small change in sphericity was observed, which indicates a slight change in pore geometry after the samples were stretched (Figure 8B).

Fig. 8

The above charts refer to the entire range of the registered pores, but do not fully reveal the changes taking place for the largest pores. This is due to the significant share of newly registered pores with small diameters. Due to the limitations associated with the resolution of the XCT measurement, the shape of the smallest pores was considerably simplified, which can be inferred from their high sphericity [16]. Therefore, a further comparison was made for the pores with a minimum diameter of 20 voxels, which corresponds to a diameter of >200 μm. The shape comparison for the largest pores, before and after the tensile test, is shown in Figure 9. As can be seen in the figure, the shape of the pores changed the most for the samples manufactured in the horizontal direction (which can be seen from comparing Figure 9A and 9B). This is confirmed in Figure 9C, which shows the changes in the projected pore size along the force axis before and after the tensile test. In the case of the samples produced in the vertical direction, no significant changes in sphericity were observed (Figure 9D–9E). This is confirmed by the graph in Figure 9F.

Fig. 9

The quantitative assessment of sphericity is presented in Figure 10. A high reduction in the sphericity of the pores was observed for the XY samples (Figure 10A). A much smaller but still noticeable reduction in sphericity occurred for the ZX samples (Figure 10B).

Fig. 10

The shapes of the pores after the tensile test for samples XY and ZX are shown in Figure 11. The presented cross-section was made in the area close to the sample's breaking point (about a millimeter above the fracture). In this case, the colors represent the surface of the porosity projected in the ZX plane of the sample's coordinate system, which corresponds to the plane perpendicular to the direction of the force.

Fig. 11

As shown in Figure 11A, there is a reduction in the projected area of pores, which results from the formation of the neck of the sample and the elongation of the pores in the force direction. Figure 11B shows a specimen manufactured vertically; here, the pore area in the plane perpendicular to the acting force is much larger, indicating that there is a lack of pore elongation in this case.

The results of the fractographic analysis complement the observations registered in the ex-situ XCT test. In the case of the fracture of the samples manufactured horizontally, a translaminar nature of the crack was observed (Figure 12). Pores with an oval shape and a small diameter are clearly visible here. The fibrous fracture morphology is directly related to the manufacturing direction, and this failure mechanism is due to subsequent layers being located in the direction of the tensile force. With increasing deformation, the crack elongation occurs in the cold flow polyamide-12 matrix. The cause of the crack in this case was the coalescence and growth in the force direction of the pores.

Fig. 12

The samples manufactured vertically demonstrate a distinct brittle fracture, which is confirmed by the obtained mechanical results. For these samples, which are characterized by layers perpendicular to the direction of the force during the tensile test, an interlaminar crack can be observed. This is due to the separation of one layer from the other (Figure 13).

Fig. 13

In this case, pores with a large surface area and large size were noticed.