Metal additive manufacturing via laser powder bed fusion or electron beam melting has been used for several years for serial manufacturing of implants. The potential to seamlessly integrate porous volumes in orthopedic implants has provided medical design engineers a powerful tool to achieve superior implant performance. However, many opportunities to further optimize the design of these porous volumes remain unexplored or do not make it from the lab to industry. This article aims to provide a coherent view on the current use of porous volumes in additively manufactured implants and some of the most recent research in this field.
The properties of porous materials are determined by features on both the microscale and the mesoscale (scale of the pores). This mesostructure gives porous materials an extra layer of complexity but also extra possibilities compared to bulk materials. In additively manufactured orthopedic devices, porous materials (scaffolds) have mainly been used to reduce stress shielding and to improve implant stability by increased bone ingrowth.1,2 The mesostructure and performance of these scaffolds is mainly determined by their porosity, pore size, and pore architecture. These are, in turn, defined by both the design of the implant and by the manufacturing accuracy.
For optimal bone ingrowth, the porosity of the scaffold should be as high as possible, fully interconnected, and the pore size should be in the 300–600 μm range.2 High porosity and pore connectivity lead to a high permeability and a large volume for bone ingrowth. High permeability is required for transport of nutrients across the scaffold before angiogenesis is complete and blood vessels take over. A large volume for bone ingrowth enables mechanical interlocking between bone and implant, which contributes to implant stability. The optimal pore size of 300–600 μm has been determined in in vivo experiments and balances easy vascularization (larger pore size) and a higher surface area (smaller pore size), which both influence bone ingrowth. In addition to porosity and pore size, the pore shape also influences bone ingrowth.3 Research has shown, for example, that the rate of tissue generation is proportional to the curvature of the surface. The type of curvature could also have an influence as in some cases tissue growth was shown to favor concave surfaces over convex or flat ones.
The requirement for high porosity is limited by the mechanical properties of the scaffold, which generally decrease with an increase in porosity. The relation between the two again depends on the mesostructure, with a much sharper decrease in strength and stiffness for bending-dominated structures than for stretching-dominated structures.4 A necessary, but not sufficient, condition for stretching-dominated behavior of lattice structures is a nodal connectivity of 6.4 scaffolds with a more or less random mesostructure have a nodal connectivity of only 3–4 and are thus always bending-dominated.5 Figure 1 shows an example of a randomized scaffold mesostructure, generated with the 3D Systems 3DXpert™ software to mimic the natural structure of bone. Figure 2 shows two as-printed fictional hip cups with different randomized mesostructures, generated with the 3DXpert software. The use of stretching-dominated structures in implants would enable a maximization of porosity without compromising mechanical properties.
The use of laser powder bed fusion (direct metal printing, DMP) enables an unrivaled level of control over scaffold mesostructure compared to conventional production of metallic scaffolds. This has enabled researchers to produce and study structures that were impossible to manufacture before. Triply periodic minimal surfaces for example (see Figure 3) have recently received considerable attention as implant materials because their mean curvature is similar to that of trabecular bone.3 Their mechanical properties and permeability have been studied to some extent but in vivo results that demonstrate their supposedly superior bone ingrowth remain to be published.6,7 A second example of recent research enabled by DMP technology is the renewed interest in the octet truss unit cell, a stretching-dominated structure suggested by Fuller in 1956.8–11 Before the development of DMP technology, porous metals with octet truss unit cells were difficult to manufacture with pore sizes in the range required for bone regeneration. They were therefore considered mostly in building or aerospace applications.11–13 Further examples of DMP-enabled progress in implant design are the use of graded porosity structures and the location-dependent optimization of the porous architecture to avoid tensile loading of bone at any implant-bone interface location.14,15 Despite the potential to optimize the scaffold meso-structure and its performance, many commercially available implants use unit cells that are not optimized for bone ingrowth or for load-bearing. One of the reasons for this is related to manufacturing inaccuracies introduced in the scaffolds during the DMP process, which are discussed next.
The production of scaffolds by DMP is much easier than by conventional manufacturing but can still be challenging. The minimal feature size is limited to the melt pool size, which can establish an upper limit for the porosity in some cases. Horizontally oriented features suffer from dross formation and thus lead to a deviation of the geometry from the designed model.9,16 The surface roughness of the scaffolds is usually much higher than the smooth design file due to dross formation and the partial melting and attachment of powder particles in the open pore space.17 The scan vectors that make up a strut are usually very short and the production process parameters (laser power, scan speed, etc.) should be adapted to this point-like exposure of the powder bed to achieve high strut densities.18,19 A high as-manufactured strut density is very important, as hot isostatic pressing (HIP) of scaffolds is less effective in reducing porosity than for bulk materials. This is due to the higher chance of open porosity in a scaffold, which cannot be closed by HIP.
Despite a high strut density, the experimental stiffness and strength of scaffolds are often lower than the theoretical values, especially for stretching-dominated structures.8–10 This is due to deviations of the manufactured geometry from the designed geometry, which change stretching-dominated structures into partially bending-dominated structures. The most important deviations are (i) strut waviness, (ii) strut thickness variation and (iii) strut oversizing and undersizing.9 Strut waviness is a misalignment of the center axis of the manufactured strut and the collinear designed strut. Strut thickness variation refers to the variation of the diameter of a manufactured strut along its length. Strut oversizing and undersizing involves a difference in manufactured strut diameter depending on the orientation of the strut. It was shown that the strut waviness and, in particular, the strut thickness variation reduced the mechanical properties most, while the extent of under-and oversizing controls the failure mechanism.9 In order to avoid a large mismatch between the designed and manufactured geometry, the current design of implant mesostructure is often guided by the manufacturing limitations discussed above. By adapting the design file to the manufacturing reality and optimizing the postprocessing steps, it should be possible to eliminate some of the manufacturing inaccuracies and raise the mechanical properties of scaffolds closer to their theoretical potential.16–20
In order to take full advantage of the potential offered by DMP, many details from the initial design stage to the postprocessing of the manufactured implant need to be fully understood and controlled. The strong dependency of DMP on software means that the use of a fully integrated solution, from design to machine control, can facilitate successful optimization of scaffold structures. In combination with thorough process and post-processing knowledge, these tools should unlock the currently underused potential of DMP scaffolds.
This article was written by Karel Lietaert, Sr. Process Development Engineer, R&D, 3D Systems, Littleton, CO. For more information, Click Here.