Characterization of Porosity in a Laser Sintered MMCp Using X-Ray Synchrotron Phase Contrast Microtomography

Direct Laser Sintering (DSL), a technology enabling the production of dense metal components directly from 3D CAD data, was used for the first time to produce a Metal Matrix Composite (MMCp) based on Al-Si-Cu alloy in view of its application in different fields, in particular for aeronautics. The porosity of the material obtained so was investigated by using optical and electron microscopy and, in particular, X-ray computed microtomography techniques. DSL is a unique technique to produce complex components in an economical way while computed microtomography is a unique technique to evaluate the porosity and pore and cracks distribution in a not destructive way. A near homogeneous distribution of the porosity and pore sizes was observed both comparing different regions of the same specimen and also by comparing different samples obtained by using the same DLS production method. A quantitative analysis of the damage in the composite is also reported.


Introduction
Direct Laser Sintering (DLS) is a technology enabling to produce dense metal components, directly from 3D CAD data.The quality of the product is comparable to that from a good investment casting, while the mechanical properties are comparable to those of a cast or machined part [1].
The application field of DLS is very wide as the so-obtained components can be used in place of almost any conventionally manufactured part, either machined or cast.In addition, the greater component complexity and feature richness, the greater the economy in production.Current applications concern subjects so different as Formula 1, Aerospace, Medical devices and Tooling.In particular, DLS technology can be used in Formula 1 and Aerospace industry not only to produce complex and structurally challenging parts made of various materials, but also it can offer lead times unachievable with other methods.
DLS enables the production of very strong but light-weight components, also containing hollows or intelligent internal structures.DLS is also unique to combine different parts into a single one, thus obtaining complex components, making them lighter and potentially improving their functionality and strength.
With the recent development of new three-dimensional (3D) characterization tools, a clear, accurate, and quantitative analysis of MMCs can be obtained.Several techniques have been used for visualization of microstructures in 3D.Serial sectioning techniques using mechanical polishing coupled with optical microscopy [14,15] or focused ion beam (FIB) milling [16][17][18] and image reconstruction have been used.
X-ray microtomography (microCT) is an excellent technique that eliminates destructive cross-sectioning, and allows for superior resolution and image quality with minimal sample preparation [19,20].
MicroCT is a 3D radiographic imaging technique, similar to conventional CT tomography systems used in medical and industrial applications.Unlike such systems, which typically have a maximum spatial resolution of about 1 mm, micro-CT is capable of achieving a spatial resolution close to 1 micron.In both conventional tomography and microtomography, hundreds of 2D projection radiographs are taken of a specimen at many different angles.The information contained in a radiograph is a projection of the absorption density inside the sample onto a plane perpendicular to the direction of the X-ray beam.If the sample is then imaged several times in different orientations, a 3D (volume) information on the sample structure can be obtained using computer algorithms.This process, referred to as image reconstruction, allows slices of the investigated object to be observed without physically cutting it.However, the maximum power of an X-ray laboratory source is limited, with a consequent upper limit to the available X-ray flux.
The micro-CT using X-ray synchrotron radiation exploits the same idea as conventional computer tomography, but several advantages come from the use of synchrotron radiation.In particular, synchrotron radiation offers the possibility of selecting X-rays with a small energy bandwidth from a wide and continuous energy spectrum and, at the same time, it guarantees a high enough photon flux for efficient imaging [21][22][23].Moreover, the use of synchrotron radiation allows for tuning the selected photon energy, in order to optimize the contrast of the different phases in the investigated samples.This possibility is of great interest for micro-CT since it allows high spatial resolution images (from 10 microns to 1 micron) to be generated, with high signal-to-noise ratio [24][25][26].
3D visualization and quantification of heterogeneous microstructures by X-ray tomography has been successfully performed in Sn-rich alloys [27], powder metallurgy steels [28], metal matrix composites [29][30][31][32], and aluminum and copper alloys [33,34].In addition to visualization, such microstructural data sets can be incorpo-rated into finite element models to predict the onset of local damage mechanisms and the macroscopic deformation behavior [32,[35][36][37].A sound understanding of damage in MMCs requires adequate visualization and quantification of fracture based on the 3D microstructure.More importantly, it requires a significant amount of statistical characterization and analysis, particularly of particle fracture and void growth.The distribution of defects and voids, before and after deformation, needs to be quantified.
Aim of the present work is to determine if DLS technique is suitable to produce an MMC door stop fitting for aeronautical application, in order to replace the up-tonow used forged titanium.Good mechanical properties and a good corrosion resistance are required for such applications.The microstructure and, in particular, the porosity play a fundamental role for those aims [38][39][40][41].
It is evident that porosity has a very strong effect on the properties of the composites.In particular, Young's modulus, flexural strength and thermal conductivity increase with decreasing porosity [38,39].Porosity appears to be detrimental for crack propagation.
Our attention was thus focused mainly on porosity: a volumetric imaging technique to investigate the samples was chosen and a quantitative analysis of the porosity has been carried out by using the VGStudio MAX Software [42].
The considered alloy is Al6061-7Si-5.6Cu 2 O.Its high density is obtained after an exothermal reaction between Cu oxide and Al particles, a suitable microstructure and a good strengthening are also obtained.

Samples
The characteristics of the materials used in the sample preparation [11] are summarized in Table 1.
The Al6061-7Si-5.6Cu 2 O alloy was prepared by blending aluminium and silicon powders in the proper mass fraction in a 2F Turbula shaker/mixer (Basel, Switzerland).DLS was performed in a commercial M250X tend machine (Electro Optical Systems GmbH, Munich, Germany) under an argon atmosphere.A CO 2 laser intensity of 8.6 W•mm −2 , a layer thickness of 0.1 mm and a hatch spacing of 0.3 mm were used.Tensile specimens of 50mm length and 3mm thickness were prepared (Figure 1).They are referred to as IFAM1, IFAM2, •••.The density of sintered materials was determined according to ISO Standard 5017.
The microstructure has been studied by optical and scanning electron microscopy (FESEM, Zeiss, Germany).For those observations, the sintered samples were sectioned along the thickness and perpendicular to the scanning line.
The phase analysis was performed by X-ray diffraction (XRD, Siemens, Germany) using Cu K  radiation.
Figure 2 shows two examples of surface morphology.Pores and metal agglomerates are visible, the agglomerates being larger than 100 m in size.
Figure 3 shows an example of XRD pattern obtained from Al-7Si-5.6Cu 2 O together with the corresponding phase recognition.

X-Ray Computed Microtomography
MicroCT experiments were performed at ELETTRA (Trieste-Italy) on beamline SYRMEP (Figure 4) with a monochromatic beam energy of 28 keV and a sample-to-detector distance of 5 cm.A two-dimensional (2D) detector recorded projections of the sample at different angular positions.In the present study 900 projections were considered within an angular range of 180°.Reference images without sample have been recorded in order to eliminate the effects of a non-homogeneous intensity or spectral composition of the X-ray beam.The exposure time was 18 seconds per projection.Images were recorded on a 2048 × 2024 CCD detector with the pixel size set to 4.5 μm.The 3D structure was finally reconstructed from 900 projections using an algorithm implemented at ELETTRA.A volume of interest was reconstructed for each sample.Each voxel of the reconstructed image was cubic with a 9 μm size.The reconstructed values of the linear attenuation coefficient for the em-ployed X-ray energy ranged between 0 and 9 cm −1 and were distributed in 256 gray levels.
Volume rendering is a 3D visualisation method wherein the data volume is rendered directly without decomposing it into geometric primitives.A 2 GHz Pentium with 1 Gb RAM and commercial software VGStudio MAX 2.1 were used to generate 3D images and to illustrate the distribution of phases in 3D.In order to achieve optimal settings for the image quality, we used Scatter HQ algorithm with oversampling factor of 5.0 and activated color rendering.

Extraction of Quantitative Parameters
Quantitative analysis of the 3D architecture can be obtained, based on the structural indices usually considered for bone samples [44].A material volume (MV) corre-   sponds to the number of voxels with an absorption coefficient in the range defined for a given material.The total volume (TV) is related to the number of voxels corresponding to all the materials in the data set in question.
The ratio of a material surface (MS) to the material volume (MV) is approximated using the Cauchy-Crofton theorem from differential geometry (it is generally not possible to calculate the material surface from polygons): the mean number of crossings per unit length of randomly chosen lines through a 3-dimensional structure approaches half of the true ratio of surface to volume [45].
Porosity is defined as the percentage of void space in a solid [46] and it is a morphological property independent of the material.In our case, the porosity depends on the fabrication process and it is an important parameter in determining the mechanical properties of the component.The total porosity (P) was obtained according to the equation: The specific surface available for pore adhesion is given by the material surface-to-volume ratio (MS/MV).3D images also enable the direct assessment of metric indices of feature sizes by actually measuring distances in the 3D space.Pore wall thickness (P.Th), and pore separation (P.Sp) or pore diameter can be thus computed.
The pore wall thickness depends on MS/MV and is calculated as P.Th = 2/(MS/MV).The mean number of pores per unit length (P.N) depends on MV/TV and is calculated as P.N = (MV/TV)/P.Th.The mean space (P.Sp) between pores depends on P.N and P.Th.It is calculated as P.Sp = (1/P.N) − P.Th.

Microstructural Characterisation
Three of the samples of Figure 1 were investigated by using X-ray Computed Microtomography.An example of the macroporous network and the microstructure of the investigated samples is reported in Figure 5.No significant differences existed in the internal microstructure among these samples.
3D quantitative parameters were calculated directly from 3D images to characterize the samples.This quantification first required segmenting the different phases to separate them from the background.In these samples, such segmentation was easily performed by simple thresholding because the gray level histogram was clearly bimodal with a first peak corresponding to background and a second peak corresponding to the material of sample (Figure 6).
The main parameters obtained from those images are summarised in Table 2.

Pore Analysis
A detailed examination of the porosity was carried out by using the spatial computational analysis techniques.The porosity computation was based on the calculation of a local pore map and was implemented using 3D Chamfer discrete distance [42].The pore value is the diameter of the largest sphere completely included in a given portion of each the investigated samples.Figure 7 shows an example of central slices in the orthogonal planes (xy, yz and xz) and a 3D volume reconstruction of the local pore sizes.Different thicknesses are plotted with different colors.After obtaining these maps, a quantitative analysis of the pore size distribution in the 3D microstructure was possible.
In Figure 7, the presence of several cracks with dimension in a range between 100 and 700 µm can be observed in addition to pores with diameters in a range be-tween 9 and 100 µm.
Figure 8 shows the distribution of 3D pore size: these distributions were obtained by dividing the thickness    map in steps of 10 µm.
From Figure 8 a large amount of micropores with radius 0 < R < 20 μm is obtained.Lower amounts are ob-served at higher radius.
The most of the radii are smaller than 55 μm, thus indicating that the sample porosity is mainly related to small pores.
In order to assess the uniformity of the distribution, the percentages of holes with different sizes were considered in three regions of the sample IFAM2.The results are shown in Table 3, which reports the percent number of pores in the three most important ranges of radius: <10 μm, 10 -20 μm, 20 -30 μm.
The distribution of pores size is uniform within the sample: in fact for each of the three intervals, the percentage does not vary much from one area (down, centre, top) to another.
The results reported in detail for sample IFAM2 are also representative of those obtained from the other two investigated samples.
In this section we also report a quantitative analysis of damage in the composite.In particular, we have analyzed the void growth in the Al-Si-Cu alloy matrix in three dimensions.Figure 9 shows the distribution of 3D cracks size: these distributions were obtained by dividing the thickness map in steps of 50 µm.
The distribution of the cracks is not the same in all the samples.In particular, the sample IFAM1 contains cracks with larger dimensions.

Conclusions
For the first time, DLS has been used for the Al-Si-Cu MMCp alloy.Porosity is an important parameter that influences the mechanical properties of the material, also in view of its application as a door stop fitting for aeronautical needs.
The results show that the porosity is distributed homogeneously inside the samples both considering its global value and also considering the contributions coming from the different classes of voids of different sizes.In particular, a porosity of 25% ± 5% was obtained for all samples.
The most of the pores have diameter not larger than 100 μm and the porosity is mostly due to a great quantity of small pores, with radius between 10 and 20 μm.
The occurrence of cracks and their not homogeneous distribution in samples nominally submitted to the same treatments can indicate that the so far obtained structure is brittle.

Figure 5 .
Figure 5. 3D reconstruction of sub-volume of sample IFAM1 showing the microstructure.

Figure 6 .
Figure 6.Gray Level Histogram of sample 1, showing the utilized threshold.

Figure 7 .
Figure 7. Example of central slices (a, b and c) in the orthogonal planes (xy, yz and xz) and 3D volume (d) of the local pore size.The different thicknesses of pores are plotted with different colors.

Figure 8 .
Figure 8. Pore size distribution in the analyzed samples.

Figure 9 .
Figure 9. Crack size distribution in the analyzed samples.