Additively Produced Ti-6Al-4V Osteosynthesis Devices Meet the Requirements for Tensile Strength and Fatigue

Abstract

The purpose of this study was to estimate the peak stresses in a laser powder bed fusion (LPBF) additive-manufactured (AM) osteosynthesis plate during physiological loading and establish if the mechanical properties of LPBF titanium alloy were suitable for this use case. Finite element models of subject-specific osteosynthesis plates for a cohort of 28 patients were created and used to calculate the peak maximum principal stresses during physiological loading, which was estimated to be 166 MPa twelve weeks post-operatively. All specimens were LPBF additively manufactured in Ti-6Al-4V alloy. ISO compliant tests were performed for tensile and fatigue, respectively. Fatigue testing was performed for specimens that had been heat-treated only and those that had been heat-treated and polished. The Upper Yield Stress was 1012.5 ± 19.2 MPa. The fatigue limit was 227 MPa for heat-treated only specimens and increased to 286 MPa for heat-treated and polished specimens. The finite element predicted stresses were below the experimentally established limits of yield and fatigue. The tensile and fatigue properties of heat-treated LPBF Ti-6Al-4V are therefore sufficient to meet the mechanical requirements of osteosynthesis plates. Polishing is recommended to improve fatigue resistance.

1. Introduction

Additive manufacturing (AM, hereafter and otherwise referred to as 3D Printing) has led to numerous innovations in orthopaedic surgery. These span from planning and surgical guides produced from polymers to implants printed with metal alloys [1,2,3,4,5]. AM technology has also enabled tailored regenerative solutions for controlled drug release [6]. Most AM processes are delivered through a layer-by-layer manufacturing strategy. Consequently, internal, overhanging and otherwise topologically complex geometry becomes manufacturable. In addition, the reduction of the need for complex machine programming, custom tooling and fixturing makes AM an attractive manufacturing process when production runs are short or even personalised on a part-by-part basis. As such, this family of processes has provided the opportunity for personalised custom implants, specifically where complex reconstruction is required [7,8].

Mainstream orthopaedic applications have often focused on creating implants with ingrowth surfaces for enhanced biological fixation; these have been used for acetabular (hip) [9,10] and proximal tibial (knee) [11] replacement implants. Metal AM implants have been used for revision joint replacement components [12], as the enhanced biological fixation is advantageous when dealing with limited bone stock, while primary implants are now becoming mainstream.

The metals used for additively produced implants include cobalt-chrome, tantalum and titanium alloys. The majority of contemporary AM implants are made from titanium alloys [13] and, in particular, Ti-6Al-4V ELI (ELI—extra low interstitials). This alloy is termed ASTM F136 [14] grade 23 (specifically for surgical implant applications) and conforms to the composition described in ASTM F3001–14 [15] and ISO 5832-3 [16]. Extensive studies have been performed on the biocompatibility of AM titanium alloy components [17,18,19,20,21]. These studies include both selective laser melting (SLM) and electron beam melting (EBM) manufacturing processes. For both AM processes, there is a large amount of data to support the low cytotoxicity of Ti-6Al-4V [22].

As well as joint replacement implants, AM has also been used for fracture fixation or osteosynthesis devices. Fractures are a common reason for hospital admission worldwide [23]; thus, osteosynthesis implants have considerably greater application than those designed for joint replacement. The use of AM technology was initially for the creation of planning guides for complex or non-standard cases [2,8,24,25]. Given the anatomic variability of bone morphology, AM has considerable potential for personalised osteosynthesis plates.

Osteosynthesis plates are also used for joint-preserving osteotomy surgery, particularly for knee osteoarthritis in the relatively young. Since a relatively young age at the time of surgery has been shown to be a major risk factor for early revision of a joint replacement [26,27], there is considerable clinical interest in delaying joint replacement. High tibial osteotomy (HTO) is a well-established procedure for treating knee osteoarthritis [28].

There is little published information on the performance of AM Ti-6Al-4V for orthopaedic osteosynthesis devices. Whereas acetabular and tibial components are principally loaded in compression [29,30], osteosynthesis plates experience bending and fail by fatigue-related mechanisms [31], typically in the presence of delayed or non-union [32,33]. The loading of HTO plates is complex and dependent upon the physiological activity being considered and the healing stage of the osteotomy. AM HTO plates provide an opportunity to provide personalised solutions for patients, which requires establishing that the AM material has appropriate properties for this use case. The key objectives of this study were to use finite element modelling to calculate plate loading during physiological activity and measure the mechanical properties of SLM-manufactured medical-grade Ti-6Al-4V ELI specimens (conforming to ASTM B348 grade 23) by a series of tensile and fatigue tests. Manufacturing defects in AM-produced parts can lead to a reduction in fatigue performance [34], as fatigue is an important failure mechanism for this type of device [35,36]. The material test specimens were inspected using SEM and nanoindention.

2. Materials and Methods

2.1. Finite Element (FE) Modelling

Computed Tomography (CT) scan data of the knee and proximal tibia from 28 subjects with knee osteoarthritis (OA) were used to create a set of proximal tibial geometries to form a cohort for an in silico trial comparing personalised with generic HTO procedures using the difference in maximum Von Mises stress in the HTO plates as the primary outcome, the full details are provided by MacLeod et al. [37]. The mean age of the subjects was 68 years (range 50 to 87 years), 54% were female, the mean body weight was 90.1 kg (range 68.8 to 121.4 kg), mean height was 1.69 m (range 1.47 to 1.90 m). For the current study, we focused on the principal stresses determined by the finite element analysis, as tensile stresses play a key role in crack propagation leading to fatigue failure under cyclic loading [38]. To facilitate interpretation of the FE results Figure A1 in the Appendix provides an overview of the whole model, exemplar physiological loading and the boundary condition, the variation in mesh density used for the various parts of the model and the locations where contact between the plates and screws was modelled.

2.1.1. Geometries and Meshing

For each subject, the CT data were segmented to obtain the proximal tibial geometry (ScanIP version M-2017.06, Synopsys, Sunnyvale, CA, USA). An open wedge high tibial osteotomy was created such that the modified mechanical axis passed through a point 62.5% [39] of the distance from the medial to lateral aspects of the tibial plateau (ANSYS SpaceClaim R18.2, ANSYS Inc., Canonsburg, PA, USA). A subject-specific plate geometry designed to match the surface profile of the subject’s tibia (TOKA, Orthoscape, 3D Metal Printing Ltd, Bath, UK) was created using specialised surgical planning software (Renishaw plc, Wotton-Under-Edge, UK). Screws were modelled as cylinders with a shaft outer diameter of 5.0 mm. The plate and screws were considered to be made from Ti-6Al-4V ELI alloy.

Experimentally validated meshing parameters were used [33]. A mesh convergence study was performed [37], quadratic tetrahedral elements were used (element size 0.8 mm for plate, screws and cortical hinge, 1.4 mm for bone, 2 mm for callus, there was local refinement around plate/screw interaction to 0.3 mm).

2.1.2. Material Properties

Linear elastic isotropic material properties were used (Ti-6Al-4V: Young’s modulus or E = 119 GPa, Poisson’s ration or ν = 0.34). For the bone, CT-based heterogeneous linear material properties were calculated using BoneMat 3.2 [40,41]. Values for E for bone ranged from 350 MPa to 23 GPa. The focus of this work was on the loads in the plate, we considered it important to represent the distribution of bone material properties rather than just considering single values for the cortical and cancellous bone as the physiological loads will be transmitted to the plate via the screw/bone interface The Young’s modulus of the callus varied with healing stage [42,43], four stages were considered: Healing Stage 1 immediately post-operation with no callus formed, Healing Stage 2 (HS2) representing 2 weeks with callus E = 1.4 MPa, Healing Stage 3 (HS3) representing 6 weeks with callus E = 24 MPa and Healing Stage 4 (HS4) representing 12 weeks with callus E = 528 MPa. As clinical rehabilitation protocols limit the weight bearing immediately post-operation, Healing Stage 1 was not modelled; the finite element analysis was performed for HS2, HS3 and HS4. Poisson’s ratio was taken as ν = 0.3 for the bone and the callus during healing stages 2 to 4 [42,43].

2.1.3. Contact Interactions

As the interest of the study was primarily plate stress, the screw-bone interaction was modelled as tied for simplification [44]. The screw-plate interface was modelled with normal contact stiffness of 0.002 and a standard Coulomb friction model with a coefficient of friction equal to 0.8 for the tangential behaviour.

2.1.4. Loading & Boundary Conditions

Five instances in each of three physiological activities (load-steps 1 to 5: fast walking, load-steps 6 to 10: chair rise and load-steps 11 to 15: squat) were modelled, with muscle and joint loads calculated using a musculoskeletal model [45] and scaled by each subject’s body weight and registered to their anatomy. The most distal part of the tibial shaft was constrained in all directions, and no other restraint was applied.

2.1.5. Solver/Solution

The FE models were solved using geometric non-linearity with an implicit solver (Ansys 18.2). The main output variables of interest in the current study were the stresses in the plate, the von Mises (σ_v), the maximum (σ₁) and minimum (σ₃) principal stresses were extracted. The peak values of these stresses over the whole cohort and all loading steps were calculated for each healing stage, HS2, HS3 and HS4.

2.2. Production of Tensile Test Specimens

The powder used to manufacture all test specimens was Ti-6Al-4V ELI (supplied by Renishaw plc, Wotton-under-Edge, UK) conforming to ASTM F136 [14] grade 23 and ASTM F3001–14 [15]. The supplier’s specification gave the powder particle size as between 15 μm and 45 μm [46]. The test specimens were additively manufactured using the SLM process using an AM250 machine (Renishaw plc, Wotton-under-Edge, UK). Laser power was set to 200 W and the layer thickness to 40 µm. The following laser parameters were used: 80 µm point distance and 50 µs exposure time for borders; 60 µm point distance and 50 µs exposure time for fill hatching. The tensile testing specimens were positioned vertically on the build plate so that the potentially weaker build direction (Z) would be tested.

Previous work has raised concerns about residual stresses in AM-produced Ti-6Al-4V components [47]. To alleviate residual stress, the specimens were heat-treated in an inert argon environment, heated to 350 °C over 60 min and held for 30 min, followed by heating to 850 °C over 60 min and held for 60 min. Specimens were allowed to cool slowly.

2.3. Tensile Testing

The tensile test specimens (n = 10) were cylindrical with a reduced working section (Figure 1), and the design of the specimens complied with the proportions stated in ISO 6892-1:2019—Metallic materials—Tensile testing [48]. The tensile test specimen geometry was created in CAD (ANSYS SpaceClaim R18.2, ANSYS Inc., PA, USA) from which the stereolithography file for LPBF additive manufacturing was created. The diameter of the working section of each test specimen was measured (150mm Digital Caliper 0.01 mm with UCAS calibration, Mitutoyo (UK) Ltd., Andover, Hamps., UK) three times (measure M1, M2 & M3 pre-test) before testing, the average diameter was used to calculate the specimen specific cross-sectional area.

Tensile tests to specimen failure were performed to determine the Upper Yield Strength (R_eH) of the additively manufactured titanium specimens according to ISO 6892-1:2019. The testing method conformed to ISO 6892-1:2019 A2. The tests were conducted using an electromechanical testing machine (INSTRON 5900 Series, fitted with a 30 kN load cell, Instron, High Wycombe, UK). The machine applied a continuous tensile force through a predefined strain rate ($\dot{e_{Lc}}$ = 0.00025 s⁻¹ corresponding to 0.005 mm/s) as per ISO 6892-1:2019 A2. Extension was measured using an extensometer (GL50, Instron, High Wycombe, UK); extension, load and the displacement of the crosshead were recorded at 10 Hz using Bluehills software (v3, Instron, High Wycombe, UK). The maximum yield stress, R_eH, and corresponding strain, as well as the 0.2% proof strength, Rp, and Young’s Modulus, E, were determined. Note that the individual specimen cross-sectional areas were used for the stress calculations.

After the completion of each test, the specimens were again measured. The cross-section diameter was measured at the narrowest intact cross-section on the parallel length, and the final gauge length was measured using the marks from the extensometer grips.

2.4. Fatigue Test Specimens

Fatigue test specimens were also additively manufactured. These were again cylindrical with a reduced working section and made to the specification required for the fatigue testing machine used (Figure 2). The specimen geometry was again created in CAD (ANSYS SpaceClaim R18.2) from which the stereolithography file was created. Two sets of samples were produced with different finishing methods: Group 1 was heat-treated and polished (Ra = 0.32 ± 0.06 µm [49], n = 16); Group 2 were heat-treated only (Ra = 5 to 7 µm [46], n = 16). All samples were measured at the narrowest point to determine the actual diameter (repeated three times). The average (mean ± SD) diameter value for each group was 3.43 ± 0.08 mm and 3.70 ± 0.02 mm for Group 1 (polished) and Group 2, respectively, and these values were used for the calculation of bending stress.

2.5. Fatigue Testing

The tests were conducted using a Wohler-type rotating fatigue machine (020313 RT, G.T.G. Engineering Co. Limited, Loughborough, UK). This machine applies a bending moment while simultaneously revolving the sample, causing total stress reversal in the specimen (stress ratio equal to −1) every revolution. When the sample fractured completely, the machine stopped automatically. The test machine was also used to record the time that the sample began and stopped revolving. The speed of the revolutions was measured using a tachometer (Standard AT-6, wavelength 630–670 nm, Premier Farnell Ltd., Leeds, UK). The total number of revolutions to fracture was determined by multiplying the total test time by the revolution speed.

The test machine was able to apply bending moments up to 6.8 Nm (with 68 main divisions, minimum increment was 0.1 Nm). The bending stress on the specimen, σ, was obtained from the formula:

$$\sigma ={\displaystyle \frac{M}{Z}}$$

where $M$ is the applied bending moment and $Z$ is the second moment of area of the specimen at the point of failure, defined by:

$$Z={\displaystyle \frac{\pi d}{32}}$$

where $d$ is the diameter of the specimen at the point of failure, which occurs at the smallest cross-section. For each group, the mean diameter given above was used.

The minimum number of samples recommended by ISO 12107:2012 [50] is fourteen; eight for the determination of the slope and six for evaluating the fatigue limit. The number of specimens selected for the present study was 16 for each group of specimens (Groups 1 and 2), with 9 being dedicated to determination of the sloping portion of the Stress-Number of revolutions (S-N) curve. This enabled the study to detect a 13.5% probability of failure with a confidence level of 90% (α = 0.1) (Equation (3), Section 5.3, ISO 12107:2012 [50]). The values of bending moment and stress chosen for the sloping portion of the S-N curve are shown in

Table 1. Values of bending moment, resulting bending stress and number of samples selected to determine the sloping portion of the S-N curve.

Bending Moment (Nm)	Group 1 Bending Stress (MPa)	Group 2 Bending Stress (MPa)	Number of Repeats
2.2	557.3	442.0	3
1.9	481.3	381.8	3
1.6	405.3	321.5	3

. To assess repeatability, testing was repeated three times for bending moments of 2.2 Nm, 1.9 Nm and 1.6 Nm corresponding to stress levels shown in Table 1.

The remaining seven specimens from each group were utilised for evaluating the fatigue strength with the staircase method (ISO 12107:2003), beginning with 1.1 Nm (determined from pilot tests) and adjusting in increments of 0.1 Nm; if a specimen failed, the applied stress was reduced by this increment for the next specimen, in case of non-failure (run out) the applied stress was increased by this increment for the next specimen. Fatigue tests which exceeded 1 million cycles were considered as non-failures or run outs. The fatigue strengths and standard deviations were calculated as per ISO 12107:2003. In this method, the fatigue strength is obtained by averaging the test stresses beyond the first test.

2.6. Microscopy

Two of each of the tensile and fatigue test specimens were sectioned, polished and etched as per Vander Voort [51]. These were imaged using a digital microscope (VHX-6000, Keyence Ltd., Milton Keynes, UK).

2.7. Nano-Indentation

Two of the tensile test specimens were sectioned lengthwise and prepared for nano-indentation. The instrumented indentation tests were performed using a Anton Paar NHT³ (Anton Paar GmbH, Graz, Austria) nano indenter using a Berkovich indenter. A maximum load of 50 mN was applied at a loading and unloading rate of 60 mN/m with a dwell time of 15 s. The indentation load and deformation depth were monitored to assess the hardness across the samples. Oliver-Pharr model was used to calculate the material hardness for each indentation. Nano-indentation was performed along the centre-line of the sectioned surface up to a distance of 5 mm from the fractured surface with 100 µm intervals to identify any variation in material and potential strain hardening during tensile tests.

2.8. Scanning Electron Microscopy (SEM)

Samples from the tensile test specimens were sliced, mounted in resin and polished for metallurgical analysis. The samples were etched using a hydrochloric/sulphuric acid mixture following the method of Vander Voort [51] to reveal the microstructure. The samples were analysed using a SU3900 Scanning Electron Microscope (SEM, Hitachi HighTech, Krefeld, Germany) with an Ultimax Energy-dispersive X-ray spectroscopy (EDX, Oxford Instruments, Abingdon, UK).

3. Results

3.1. Finite Element Modelling

The peak values of maximum principal stress generally occurred at the contact between the plate and the screws (Figure 3), quite often at the screw hole immediately above the bridging span of the plate.

The effect of the healing stage on the predicted plate stresses was substantial; the peak values of the stresses are plotted as box and whisker plots for the whole cohort for each load step in Figure 4. Over the cohort, there was considerable variation in the peak values of Von Mises, maximum and minimum principal stresses for each loading step, with a clear reduction in stress values as healing progressed from HS2 to HS4.

The peak values over the whole cohort and all loading steps for each healing stage are given in Figure 5; the peak values for maximum principal stress (σ₁) drop from 872 MPa for HS2, to 230 MPa for HS3 and 166 MPa for HS4. Peak values of Von Mises stress (σ_v) (HS2: 1423 MPa, HS3: 545 MPa, HS4: 165 MPa) were generally higher than the peak maximum principal stress, reflecting the importance of the minimum principal stress. The lowest minimum principal stress values were −1704 MPa for HS2, −546 MPa for HS3 and −186 MPa for HS4.

3.2. Tensile Tests

The pre-test measurements of tensile test specimen working section diameter ranged from 3.45 to 3.50 mm, mean (±standard deviation or SD) value was 3.46 ± 0.02 mm (

Table A1. Reduction in working section diameter as a result of testing for each of the tensile test specimens.

Specimen	Pre Average Diameter (mm)	Post Average Diameter (mm)	Difference (mm)	Percentage Difference (%)
1	3.46	3.28	−0.17	−5.01
2	3.47	3.25	−0.22	−6.35
3	3.45	3.27	−0.18	−5.31
4	3.50	3.35	−0.15	−4.29
5	3.46	3.31	−0.15	−4.43
6	3.46	3.27	−0.19	−5.49
7	3.46	3.25	−0.21	−6.17
8	3.44	3.24	−0.20	−5.72
9	3.45	3.30	−0.14	−4.16
10	3.45	3.31	−0.14	−4.06
Average	3.46	3.28	−0.18	−5.10

).

A representative stress-strain curve from the tensile testing is given in Figure 6. All specimens exhibited a similar behaviour, with stress increasing linearly with strain until yielding occurred, resulting in necking and ultimately failure.

The mean value, over all tensile test specimens, of the peak force measured was 9518.8 ± 119.8 N, and the corresponding mean R_eH was 1012.5 ± 19.2 MPa (

Table A2. Peak force, peak stress and Young’s Modulus from the tensile testing.

Specimen	Peak Force (MPa)	Peak Stress (MPa)	Youngs Modulus, E (MPa)	Initial Gauge Length (mm)	Maximum Extension (mm)	Maximum Extension (%)
1	9551	1015	112,679	19.90	2.74	13.78
2	9563	1011	107,801	20.50	2.92	14.23
3	9373	1003	110,616	21.49	3.09	14.38
4	9353	972	112,737	21.64	2.48	11.47
5	9382	998	111,863	20.91	2.76	13.20
6	9624	1024	111,321	21.50	2.54	11.80
7	9423	1002	108,416	20.75	3.20	15.41
8	9696	1043	113,196	21.02	3.04	14.47
9	9566	1023	110,096	20.87	3.00	14.38
10	9656	1033	112,079	21.55	2.94	13.66
Mean	9519	1012	111,080	21.01	2.87	13.68
SD	120	19	1746	0.52	0.22	1.16

). The mean value of the 0.2% proof stress, Rp, was 911.6 ± 21.5 MPa. The mean Young’s Modulus was 111.1 ± 1.7 GPa. The mean specimen minimum diameter post-test was 3.28 ± 0.03 mm, equating to a 5.10% reduction in diameter or 9.96% reduction in cross-sectional area. The average elongation at fracture across all 10 tensile specimens was 13.68 ±1.16% (Table A2).

3.3. Fatigue Tests

For the determination of the S-N curve gradient, the number of cycles to failure was more variable in the polished specimens (Group 1,

Table A3. Fatigue test results for the slope of the S-N curve, showing cycles to failure for a given applied stress, Group 1 heat treated and polished, Group 2 heat treated only.

		Number of Cycles to Failure
		Repetition Number
Group	Stress (MPa)	1	2	3	Mean	SD
1	557	27,351	36,468	30,390	31,403	4642
	481	36,468	30,390	54,702	40,520	12,652
	405	45,585	45,585	33,429	41,533	7018
2	442	33,429	27,351	24,312	28,364	4642
	382	48,624	51,663	42,546	47,611	4642
	321	100,287	85,092	79,014	88,131	10,957

). For the polished, heat-treated specimens (Group 1, Figure 7), the most probable estimate of the mean S-N relationship was:

$${\widehat{\mu }}_x=6.798-0.00464 y$$

where ${\widehat{\mu }}_x$ is the log of the estimated number of cycles for a given stress y. Similarly, for the heat-treated only specimens (Group 2, Figure 7), the most probable estimate of the mean S-N relationship was:

$${\widehat{\mu }}_x=7.526-0.00743 y$$

The sequential progress of the staircase method for both groups is plotted in Figure 8. The fatigue strength for Group 1 (heat-treated and polished specimens) was 285.9 ± 10.3 MPa, and that for Group 2 (heat-treated only) was 226.8 ± 8.2 MPa.

3.4. Microstructural Analysis

Figure 9 shows a representative microstructure of the LPBF Ti-6Al-4V samples after polishing and etching; all specimens had a similar appearance. The range of microstructures formed during AM has been overviewed by Tong et al. [52] and examined by Vrancken et al. [53]. The Ti–6Al–4V alloy examined in this study is an α–β alloy, where the α-phase has a hexagonal close-packed structure, and the β-phase has a body-centred cubic structure. During additive manufacturing, the heating and cooling rates can be extremely high, and can lead to the formation of a non-equilibrium martensitic α′-phase, rather than equilibrium α– and β-phases, which can increase the strength but is at the expense of reducing the ductility. A heat treatment has therefore been employed here to enable the initial martensitic α′ to transform to α and β. As an example, Vranken et al. [53] reported that after a similar heat treatment of 850 °C for 2 h on SLM parts, the fully martensitic α′-phase structure of a Ti–6Al–4V was converted to a mixture of α and β, where the α-phase was present as needles. It can be seen from Figure 9 that the microstructure exhibits a similar needle-like structure as a result of the conversion of the α′ martensite phase to α and β, and such a microstructure was considered to produce to best optimum mechanical properties in terms of strength and ductility [53].

3.5. Nano-Indentation

Instrumented indentation was used to characterise the nano hardness of the sample as the distance increased from the fractured surface. Figure 10 illustrates the plastic deformation of the sample as the load was applied using a Berkovich indenter. The graph shows a clear distinction between the material behaviour near the fractured surface to that of the bulk material at 2.5 mm and 5 mm distance from the surface. The Oliver-Pharr method was used to calculate the nano hardness at various points from the force-plastic deformation results. The measured hardness of the tensile sample at 100 μm intervals from the fractured surface is shown in Figure 11. The sample has a bulk hardness of 429 HV. As shown in Figure 11, as the distance from the fractured surface increases, the hardness drops to reach the bulk material hardness of ~413 VH. The increased nano hardness closer to the fractured area can be attributed to strain hardening during tensile testing.

3.6. Scanning Electron Microscopy (SEM)

The SEM micrograph of the fractured edge is shown in Figure 12. Whilst there is an indication of microstructural alignment as a result of tensile loading, the failure appears to have been initiated from the pores within the material. The presence of porosity and semi-molten powder particles in laser powder bed fusion samples is a well-known phenomenon, and typically, the density of such samples is below 99%. These can act as stress concentration points and weaken the material compared to fully dense wrought alloys. Figure 13 illustrates a high magnification micrograph of a pore.

EDX analysis of the samples confirms the composition of the material mainly consisting of titanium (Ti), aluminium (Al) and vanadium (V) with trace amounts of nitrogen (N) and iron (Fe) with zirconium (Zr) present in the ceramic resin used for mounting the sample and Silicon (Si) associated with the polishing compound. The latter two elements were only detected around the edges of the sample. Figure 14 demonstrates the EDX spectrum colour maps of the main constituent elements of the sample. The analysis clearly showed that the main elements (Ti, Al and V) are evenly and randomly distributed across the sample with no areas of high concentration. Repeating the analysis at the centre of the sample, away from the edges, confirms the composition with an average of 6.3 wt% Al, 4.38 wt% V and balance Ti as shown in the spectrum analysis in Figure 15.

4. Discussion

The purpose of this study was to estimate the peak stresses in an AM osteosynthesis plate during physiological loading and establish if the mechanical properties of LPBF titanium alloy were suitable for this use case. The finite element modelling predicted peak tensile stresses of approximately 872 MPa during physiological loading at two weeks post-operatively over a cohort of 28 subjects with body weight in the range 68.8 to 121.4 kg; this peak value dropped to 230 MPa at six weeks and 166 MPa at twelve weeks post-operatively.

A series of tensile and fatigue tests were performed on additively manufactured specimens. Comparing the tensile test results with the standard industry values [46,54] and published values [55], the findings were within the normal range for the alloy. For additive manufacture, the vertical direction has lower values compared to the horizontal direction for Young’s Modulus and ultimate tensile strength [46]. This study found that the Young’s modulus (111.1 GPa) was at the lower end of the additive machine manufacturer’s values. The percentage elongation at breakage of the titanium alloy (13.68%) was within the reported range. The yield strength was found to be 911.6 ± 21.5 MPa, similar to that reported by Vrancken et al. [53], which was lower than the value quoted by the additive machine manufacturer but within the range quoted for the material [54]. The values for yield strength and upper yield strength are comparable to those quoted for yield strength and ultimate tensile strength by Alcisto et al. [55] for wrought titanium alloy, and so are the percentage elongation values. Alcisto et al. reported that whilst yield strength and ultimate tensile strength of their additively manufactured material was similar to wrought material, the percentage elongation was substantially lower. A possible explanation for the improved percentage elongation could be the combination of smaller grain size powder and higher laser power in more recent additive manufacturing machines.

The microscopy analysis of the microstructure showed the appearance typical of LPBF Ti-6Al-4V parts that have been heat-treated [56]. The indentation tests confirmed strain hardening behaviour during tensile testing. SEM analysis showed the presence of porosity, as is common for laser power bed fusion parts, with EDX analysis showing the expected composition of the titanium alloy. LPBF Ti-6Al-4V samples can typically reach up to 99% density. The presence of pores can reduce the mechanical and fatigue strength of the samples through reduction of cross cross-sectional area and/or introducing high stress concentration areas [57]. Pessard et al. [58] identified a critical defect size of 30 µm. They noted that pores with $\sqrt{area}$ larger than this threshold negatively impact the fatigue strength of LPBF Ti-6Al-4V alloy. Porosity can be controlled by selecting appropriate AM processing parameters, i.e., scanning speed and hatch distance and through postprocessing such as high-pressure isostatic pressing [59].

The LPBF Ti-6Al-4V material exceeded the minimum requirements of ISO 5832-3:2019 Implants for surgery—Metallic materials. Part 3: Wrought titanium 6-aluminium 4-vanadium alloy for proof strength (800 MPa), ultimate tensile strength (900 MPa) and percentage elongation (10%).

Patients who have received orthopaedic treatment walk approximately 1.9 million cycles a year [60]; however, during the early recovery period, partial weight bearing is normally prescribed, which significantly reduces peak ground reaction forces [61]. For lower limb osteotomies, healing (defined as bone union) occurs in 4.5 ± 1.5 months [62]. Once healing has occurred, the majority of the load is transmitted through the bone and the forces passing through the osteosynthesis device are significantly reduced. The fatigue testing reported here demonstrated the effects of surface finish, heat-treated and polished specimens had a fatigue strength of approximately 286 MPa, whilst that for specimens which were only heat-treated was approximately 227 MPa. This finding is supported by those of Hu et al. [34], who reported surface defects had a greater effect on fatigue performance than internal defects. Even without polishing, additively manufactured osteosynthesis devices exceed the functional fatigue requirements for this class of device. The finite element simulations performed in the current study used full weight-bearing forces and joint reactions. Typically, osteotomy patients are told to refrain from full weight-bearing for six weeks. Although, even when early weight-bearing is encouraged, it is likely to occur between two and five weeks following surgery [63]. Therefore, the stresses in the plate at Healing Stage 2 represent a worst-case scenario, and even given this worst-case scenario, the stress in the plate was comfortably below the yield strength. The maximum load at six weeks, when patients are more likely to be fully weight bearing, was also well below the fatigue limit of the heat-treated and polished material. Three months post-operatively, the peak loads in additively manufactured osteotomy plates were well below 180 MPa. For osteosynthesis applications using additive manufacturing with Ti-Al6-V4, we would recommend that components be heat-treated and polished.

A limitation of this study is that only specimens from one particular type of AM machine, which used the SLM process, were tested. The specimens were manually finished or polished where appropriate, which introduces variability into the manufacturing process; however, this is representative of custom-made medical devices, where each component/system is unique. The plate is custom, and therefore, the shape changes for every single patient; changes in geometry can significantly influence stresses. The finite element modelling was performed for a cohort of 28 subjects with knee OA, and therefore covered a range of geometry variations.

5. Conclusions

The tensile and fatigue properties of heat-treated LPBF Ti-6Al-4V are sufficient to meet the mechanical requirements of osteosynthesis plates. Whilst 99% density can be achieved in laser powder-bed fusion (LPBF) additive manufacturing, the presence of porosity can impact the mechanical and fatigue performance of the additively manufactured samples and act as stress concentration areas. Heat treatment and polishing of components are recommended for osteosynthesis applications.