An investigation into the properties of 3D printed Ti6Al4V FCC lattice structures with different strut thicknesses

. Metal additive manufacturing of titanium and its alloys can produce complicated geometries cost-effectively while maintaining biocompatibility. It is known that the material property differences between bone and Ti6Al4V cause stress shielding, leading to bone failure around the implant. Using lattice structures is effective at reducing elastic modulus while improving osteointegration. However, it is important first to characterise the as-printed material to investigate the effects of lattice structures on the bulk material properties. Understanding the microstructure, porosity, and related mechanical properties can discern the bulk material properties of the unit cell. The microstructure of printed samples was found to be martensitic. The printed samples contained porosity with strut thickness deviations ranging from the design from 44.29 % (t = 0.50 mm) to 28.43 % (t = 1 mm). It was found that the high amount of porosity resulted in considerable variation in compression material properties.


INTRODUCTION
Additive manufacturing (AM), instead of subtractive manufacturing techniques, combines materials to produce products from 3D model data, usually layer by layer. Powder bed fusion (PBF) is an advanced subset of AM where a high energy source is used to selectively melt or sinter metallic powder. PBF can be further categorised into techniques depending on the nature of the power source. Selective laser melting (SLM) is one of the categories of PBF. It uses a high-energy laser to selectively melt the powder, as shown in Figure 1, resulting in a fully dense product with comparable mechanical properties to conventional fabrication techniques [1]. Fig. 1. PBF printing process [2] Ti6Al4V is a high-specific-strength alpha-beta titanium alloy with exceptional corrosion resistance. It is one of the most widely used titanium alloys, and it is utilised in various applications that require low density and strong corrosion resistance, such as aerospace and biomechanical applications [3]. Typically, this titanium alloy is heat treated to vary the amounts of alpha (α) and beta (β) phases in the microstructure, which results in different mechanical properties [4]. Depending on cooling rates, different microstructures can be obtained, including Martensitic, Widmanstätten, grain boundary allotriomorphic α, or globular primary α. It is also important to note that a martensitic structure usually consists of α', which has a hexagonal (HCP) or α", which has an orthorhombic structure [5]. The SLM PBF printing process parameters significantly impact the final material properties of Ti6Al4V samples [6]. The high thermal gradients from the printing process result in increased internal stresses of final parts, negatively affecting the material properties. It is thus usually suggested that a stress relief heat treatment (SR) is applied before use [7]. The mechanical properties typically expected after stress relief are contained in Table 1. Failure to relieve internal stresses could result in components becoming distorted when removed from the base plate or failing early under fatigue conditions. It is known that the elastic modulus mismatch between bone and solid Ti6Al4V products causes stress shielding, which can lead to bone failure around the implant [9], as shown in Figure 2. It has been demonstrated that using lattice structures effectively reduces the elastic modulus of metal parts closer to the bone [10]. However, traditional manufacturing cannot produce lattice structures, and AM SLM is ideal because of the lattice structure's small size and complex geometry. Therefore, it is essential to understand the core principles and mechanisms before trying to understand the effect of lattice structures on bulk material properties. This data can help further research into the topic of reducing the elastic modulus mismatch by making use of SLM PBF lattice structures.

Fig. 2. Stress shielding indicated by SS [11]
Previous authors have studied the effect of lattice porosity and dimensional deviation on the mechanical properties of the built part [1,10,12]. In a study of Body centred cubic (BCC) structures of AlSi10Mg, dimensional deviations were found when strut thicknesses fell below 1.5 mm diameter and resulted in losses of cross-sectional areas up to 40% [12]. In another study, porosity resulted in a loss of density ranging from 0.5% to 10.31% for PBF lattice structures with curved lattice designs [12]. Diamond lattices studied showed dimensional deviations of 10% with losses in the cross-sectional area from porosity of 7.9% [1]. Dimensional deviations were attributed to differences in volume and border laser strategies [12], struts were found to be larger than design due to sintering of powder on the outer surfaces when high energy inputs were used [12], and porosity resulted in compressive strength a loss of 31 MPa from simulation models [1] This study aims to understand the effects lattice strut thicknesses have on porosity and dimensional deviation with the view of helping future studies produce better structures suited for osteointegration. In this paper, three different lattice strut thicknesses with a face-centred cubic (FCC) unit cell structure were built through SLM PBF using Ti6Al4V. These were investigated to characterise the mechanical properties and phases of the struts and solid printed structures.

Design and Printing
Three different lattice strut thicknesses of an FCC unit cell structure were produced with SLM-PBF. An FCC structure was chosen because it is a relatively well-known and simple structure. It is also a self-supporting structure which simplifies the printing process. Throughout this document, t indicates strut thickness. The lattice designs were created on nTopology CAD software with a preview in Figure 3. The samples were produced on a Renishaw AM400 with a 1070 nm laser, 65 μm spot size and a point-by-point laser exposure methodology as stipulated in Table 2. Samples were manufactured with strut thicknesses of 0.5, 0.75 and 1.0mm. An additional completely solid sample was printed for comparison. Unit cells were 3mm in size, with the entire volume being 15mm X 15mm and the solid specimen 10mm X 10mm. The final products and the close-up of the general surface of the lattices are found in Figures 4 and 5, respectively.

Stress relief heat treatment (SR)
The samples were put through a stress-relieving heat treatment to ensure all the residual stresses were relieved following SAE-AMS-H81200 [13]. They were heated at a rate of 200°C/hour, kept at 650°C for four hours, and furnace cooled, as shown in Figure 6. The SR was conducted in a vacuum furnace to avoid oxidation.

Metallographic analysis
Metallographic analysis helps in analysing the microstructure of the samples. It also aids in showing any imperfections, porosity, and other defects that will directly affect the mechanical properties of the Ti6Al4V printed alloy. The samples were mounted, ground, and polished, after which they were etched with a Kroll etchant following ASTM E407-07 to reveal the microstructure [14]. It is important to note that all micrographs were taken via optical microscopy and perpendicular to the build direction.

Computerised tomography (CT) scanning
Similar to the metallographic analysis, CT scanning was used to view porosities and defects that will directly affect the mechanical properties. The CT scans were conducted with a Nanotom S nano CT scanner.  Fig. 7. Nanotom CT scanner [15]

Indentation Plastometry (PIP)
Profilometry-based indentation plastometry (PIP) is becoming more common in mechanical testing methods. Like a hardness test, a small indent is formed and measured, and then the indentation process is subjected to an iterative finite element method (FEM) simulation of the indentation process to produce load-displacement plots, true stress and strain curves, and hardness values for the specific material. [16].

Metallographic analysis
From the metallographic analysis in Figure 11, the microstructure for all sections was similar for all samples. The microstructure consisted of a fine acicular α' martensitic phase (Figure 9 below) with similar microstructures for the different lattices and the solid printed block. The microstructure is a function of the print parameters. There was porosity in the printed lattices and the solid block ( Figure 10). Porosity present is a function of the printing process parameters [10]. An increase in porosity is expected with an increase in scan speed, hatch spacing, and a decrease in laser power [6]. Porosities result in a reduction in cross-sectional area. However, surface porosities are generally desirable in the medical implant field as they result in effective osteointegration [1]. The porosity was randomly distributed and present at varying degrees within the samples. It is worth noting that the printing parameters of the border and the rest of the volume are usually different, as shown in Table 2. Through metallographic analysis, there was some dimensional deviation between the strut design diameter and the actual strut diameter; this aligns with the literature. The micrographs of the midway of the strut thickness were taken, and ImageJ software was used to measure the strut thicknesses. A summary of the average strut thickness and the average pore diameter measured is contained in Table 3.  It is important to note that all the dimensional measurements above are affected by how accurate the midway cross-section during sample prep was achieved. The total reduction in cross-sectional area can be calculated assuming a reduced strut thickness from measured data and the presence of a pore in this strut, and the results are summarised in Table 4.

34.03
The results from the above table should be noted as the worst-case scenario, as the lattice strut would not be larger than midway. The results also assume that porosity would be present 370, 08002 (2022) https://doi.org/10.1051/matecconf/202237008002 MATEC Web of Conferences 2022 RAPDASA-RobMech-PRASA-CoSAAMI Conference at the reduced strut, but this is not always the case. However, it is noted that it will reduce the structural integrity of the lattice, and a means to address this would allow for a more reliant structure overall. The results calculated follow Vrana [12] trends, where losses in the cross-sectional area are found when the lattice strut thicknesses are reduced.

Computerised tomography (CT) scanning
The CT scanning provided a view of the porosity distributed within the material in Figures  12 and 13. The CT scans in Figure 12 clearly show that the lattice designs' porosity was relatively evenly distributed. It is, however, apparent that the solid build sample in Figure 13 had less porosity, and porosity was mainly located on the edges, consistent with the literature [12]. Note that results show the outer surface on the left and the porosity with its size on the right.  As seen in Table 5, the t=0.50mm and t=0.75mm had a very similar porosity volume ratio of around 1.5%. The t=1.00mm and the solid sample have a lower porosity volume ratio. The average porosity size increases slightly with increased strut thickness. The reduction in the cross-sectional area of struts in Table 5 assumes a complete pore in a strut. The CT scans were also used to get a more accurate analysis of the dimensional deviations between the strut design diameter and the actual strut diameter. The results are contained in Table 6. The combined reduction in the cross-sectional area of a strut, assuming one complete pore is present and considering the dimensional deviation of diameter, results in a substantial decrease in cross-sectional area, as seen in Table 6 below. The results correlate to results by Vrana, where a reduced strut thickness was found to have a more significant dimensional deviation [12]. Their findings found this to be a 40% loss in the crossectional area, whilst deviations in this study show this loss was found in strut thicknesses of 0.5 mm only. Trends in results from metallography correlate well with CT analysis.

Indentation Plastometry (PIP)
The PIP test was conducted on the samples to obtain final representative results. The test was conducted perpendicular to the print layers (In the build direction).  All the material properties from the PIP results line up with the typical values expected and available in the literature [8] in table 1. However, from Figure 14 and the relatively large uncertainty ranges in Table 9, there was a lot of variation in the data. Scatter in results is attributed to the porosity present in the samples. Since PIP is a local deformation test, some of the indents during the test did not encounter any porosity, while others did. Zhang found the modelled loss in compressive strength to be 31 MPa [1], with the standard deviation in this analysis found at 88 MPa.

CONCLUSION
The work looked at microstructural analysis, cross-sectional area analysis through metallography and CT analysis, and PIP indentation plastometry.
The cooling rates during SLM produced an α' phase (Martensite). Metallographic and CT analysis found a reduction in the cross-sectional area from the design. The 1 mm strut thickness showed the least total cross-sectional area loss among all samples. CT scanning correlates with metallography finding a reduction in cross-sectional areas of 51.89 % and 34.03 % for the t = 0.50 mm, and t = 1.00 mm, respectively for metallographic analysis and CT scans finding 44.29 % and 28.43 % for the t = 0.50 mm, and t= 1.00 mm, respectively. The CT scans provide a more accurate representation of the actual values due to uncertainty in metallographic sample preparation.
The PIP results showed that porosities impact the local compression behaviour and hardness of SLM PBF samples, which resulted in considerable uncertainty with yield stress ranging from 990 MPa to 1163 MPa; this was attributed to the presence of pores at areas analysed with lower yield strengths.
The reduction in the expected cross-sectional area because of dimensional deviation and porosity in the lattice structures will likely decrease the expected bulk material properties. Therefore, further research will have to be done on minimising or accurately predicting the dimensional deviation and porosity.