Cleaning techniques and non-destructive verification of residue removal within additively manufactured implants

The human body is incredibly resilient, but not infallible. Musculoskeletal injuries can manifest that the human body is not able to repair or treat itself, and without treatment, can severely degrade quality of life; these injuries require surgical intervention in order to deliver an appropriate correction, and improve an individual’s quality of life and end outcome. Off-the-shelf implants have existed for many years, and though a variety of materials have been used in the past, titanium alloy (specifically Ti-6Al-4V) has become the most common implant material, due to a large body of clinical evidence, as well as good mechanical and biocompatible properties. These off-the-shelf implants are generally solid, and may require surgical adjustment during implantation, leading to increased surgical and recovery times. Additive manufacturing affords a design revolution for implant creation, allowing for aspects such as: mechanical property tailoring to match the surrounding bone and implant site more closely; allowances for patient specificity to match a patient case more precisely; and allowing for the generation of porosity, lattice, and lattice gradient design, tying in with both mechanical property tailoring, as well as allowing for bone ingrowth and angiogenesis (blood vessel formation). These aspects cannot be matched by traditional subtractive manufacturing means. Laser-based powder bed fusion (LB-PBF) is one such method of additive manufacture, using metal powder as the input material. However, the metal powder input presents potential issues for implant applications; if residual powder not joined to the implant pool during the manufacturing process is allowed to run free in the body, there is a high likelihood of harm, having the potential to cause inflammation, infection, bone resorption, and ultimately, implant failure. As such, there is a need not only to clean an implant structure to remove residual powder, but also verify its removal for patient safety. The aims of this research were to explore the following two aspects; firstly, to investigate different cleaning or residue removal techniques for powder-based AM implant structures for the goal of creating a robust and repeatable cleaning methodology; and secondly, to investigate different imaging modalities to explore viable and non-destructive methods for statistical characterisation of residual unwanted manufacturing material residue. State of the art literature has been reviewed, detailing the use of cleaning or post-processing for removal of manufacturing material residue from LB-PBF implant structures. Testing and verification methods for ensuring powder removal are also detailed. Pitfalls and gaps are also noted, including the lack of standards and robust protocols for implant cleaning, and the need for destructive testing to ensure verification of powder removal from implant structures by regulatory bodies, which is antithetical to the freedom and flexibility that additive manufacturing affords; non-destructive testing methodologies are thus described. The lack of tools and software for non-destructive analysis was an additional gap found within the current state of the art of the field. Three different imaging modalities were explored for their viability for image penetration of a dense lattice structure: digital microscopy, scanning electron microscopy and micro computed tomography (µCT). While the former two allowed greater surface detail, only µCT allowed for three-dimensional scanning and penetration. The ability to generate penetrative image data gave basis for the development and creation of an algorithmic methodology for the detection and categorisation of partially attached particle materials within LB-PBF structure. The algorithm was tested using three different µCT datasets of Ti-6Al-4V buckling struts manufactured at 30°, 60° and 90° build orientation angle respectively. Several histograms were generated, presenting categorical data for analysis, with a major takeaway between the datasets being the reduction of partially attached particles detected the higher the build inclination angle utilised. Cleaning of lattice implant structures and structural components was also conducted in order to facilitate the generation of a robust post-processing protocol. The first experiment compared ultrasonic cleaning against processing via centrifugal acceleration; while there was no significant degree of difference, several concerns were noted, including feasibility of centrifugal processing on larger scale structures, and proper sample handling and preparation to present infeasible results. The second experiment used ultrasonic cleaning on a full-scale sized tibia implant. With the addition of a surfactant, a large amount of material (over 3000mg) was removed across six, one-hour cycles. The final experiment trialled a use dry-ice blasting to remove residue from lattice structures. Three different time points were chosen, and the shortest time point proved on average to remove the most manufacturing material residue, with the largest time point presenting instances of blaster and blaster input material instability. The combination of ultrasonic cleaning and dry-ice blasting in tandem presents a robust cleaning protocol to remove manufacturing material residue from LB-PBF implant structures. Finally, µCT datasets were captured from the third experiment pre-blast and post-blast, and trialled with the particle detection algorithm in order to provide categorical information. Unfortunately, the datasets could not be processed by the algorithm in its current state; the information in full datasets was too great for the computer hardware to handle, and partial datasets while overcoming this problem had different issues present that meant they not able to be further processed for analysis. While this is a disappointing result, the particle detection algorithm in its current state represents a toolset for non-destructive testing and analysis, of which the current field presents a distinct gap and lack. Major contributions of this research and this thesis are the development of a robust cleaning protocol for additively manufactured implants, and the generation of an algorithmic methodology for analysing µCT datasets for characterisation of attaching manufacturing material residue. The former presents a method that does not leave additional particle residue and presents a multi-faceted approach for both partially melted (sintered) and embedded material through dry-ice blasting and ultrasonication. The latter is a software algorithm that can be used to compare different post processing or cleaning states (or time points) and can be used to characterise the amount of removed material as well as remaining material from post processing or cleaning operations, which is an area that the current field has not addressed.