Mechanical properties characterisation of metallic components produced by additive manufacturing using miniaturised specimens

ABSTRACT
 The various process-specific differences in techniques compared to traditional techniques can produce significantly different mechanical behaviour in additively manufactured (AM) parts compared to traditional bulk counterparts. Components produced by AM are built layer by layer via localised melting. Therefore, both location- and orientation-dependent properties can be expected. Since many AM parts take advantage of the design and topology freedom provided by AM, properties characterisation with the use of standard specimens is not always possible, requiring the use of small-sized specimen techniques. In the current paper, three AM-produced IN-718, Ti-6Al-4V and H13 parts using electron beam powder bed fusion and laser powder bed fusion are evaluated. Local mechanical properties have been assessed with the use of mini-tensile tests that were developed for cases where limited amounts of material are available. The results obtained demonstrate the ability to measure location- and orientation-dependent properties in AM components using such approaches and highlight that additional work by the AM community remains in order to determine the source(s) of such differences.


Introduction
Detailed characterisation of engineering materials and components is an important part of the design process and helps to ensure safe component use.In general, the various processing techniques used in the production of many components can produce local property variations in large components that have been produced by welding, coating techniques or additive manufacturing (AM) (Wang, Williams, Colegrove, et al. 2012;Hernandez-Nava et al. 2016;Machry et al. 2016).Electron beam powder bed fusion (EB-PBF) and laser powder bed fusion (L-PBF) systems use a high-power electron or laser beam as an energy source for powder melting, respectively (Sun et al. 2017;Antonysamy et al. 2013;Lewandowski and Seifi 2016;Seifi et al. 2017;Šafka et al. 2016;Chlebus et al. 2015;Mertová et al. 2020;Gotterbarm et al. 2020;Pehlivan et al. 2020).In L-PBF, the deposition process usually is conducted in the protective atmosphere of inert gas with little powder bed preheat while significant powder bed preheat is used in EB-PBF.This produces significantly different heating and cooling conditions between these technologies.
The current paper discusses local property characterisation for additively manufactured components via the use of miniaturised tensile tests (M-TT).This is very relevant to the design process as components produced by AM techniques can yield property variations that are both location-and orientation-dependent (Yuan et al. 2012;Džugan, Procházka, et al. 2015;Dzugan, Sibr, et al. 2017;Kohyama et al. 1991).The properties may also vary with respect to the thickness, length or angle, as well as at the contacts with the supporting structures.One source of these property differences relates to the varying heating/reheating and cooling conditions at various locations of complex parts produced by AM.Unfortunately, standard testing procedures are unable to characterise the local properties in complex-shaped objects due to the large size requirements for standard-sized specimens in existing ASTM standards.Therefore, new techniques must be established for such detailed local characterisation.The approach of using M-TT samples has been used by a variety of research groups, including ours (Yuan et al. 2012;Džugan, Procházka, et al. 2015;Dzugan, Sibr, et al. 2017;Kohyama et al. 1991;Konopik et al. 2018;Gussev et al. 2015;Wang, Wang, et al. 2013;Rund et al. 2015;Džugan, Rund, et al. 2017;Kumar et al. 2014;Kohno et al. 2000;Džugan, Konopik, et al. 2015;Dzugan, Prochazka, et al. 2017;Shin et al. 2015;Dzugan et al. 2020), although this work utilises this approach to evaluated printed parts.
The present work summarises detailed local characterisation of properties in AM parts that includes an L-PBF-processed IN-718 blade, an EB-PBF-processed Ti-6Al-4V propeller and a simplified L-PBF-processed H13 tool steel experimental propeller.M-TT specimens were excised and/or printed alongside such parts in order to document location-and orientation-dependent properties as well as confirm the suitability of using miniature specimens that can facilitate AM process optimisation.The results from such miniature specimens can also be used for component FEM model creation that includes actual local properties to enable design optimisation, although this is not investigated within the current study.Furthermore, M-TT specimens printed along with components can also be utilised as witness specimens to provide a low-cost process evaluation tool that only requires a relatively small material volume and simple testing procedure.
Local property characterisation of AM components is demonstrated here on three AM-processed objects made of IN-718, Ti-6Al-4V and H13 tool steel.The microstructures were investigated using scanning electron microscopy (SEM) and light microscopy (LM).The material integrity was studied by means of X-ray microcomputed tomography (µ-CT).In addition, the M-TT specimens after testing were subjected to SEM fractographic observations.This paper presents the initial results of M-TT excised from AM-processed parts to document differences in local mechanical properties.Additional work remains for the AM community to determine the source(s) of such differences, along with modelling efforts to eventually produce born qualified parts.

Materials and experimental procedures
Processing and specimen preparation Due to the relatively small size of many components, the application of standard mechanical testing methods may be impossible to determine critical local properties that may be different than those obtained from bulk specimens.That requires exploration of sub-sized specimens.Three different alloys, processing methods and parts have been utilised in order to demonstrate the generality of our miniaturised specimen testing approach, while also documenting significant location and orientation differences in tensile properties.tool steel powders were used to manufacture several components while the details of processing for each alloy/component are outlined below.The M-TT specimens were extracted from the AM deposited components with different specimen dimensions due to the different component geometries printed in addition to the desire to characterise the local properties (including anisotropy) in different locations/orientations of the parts.While these different specimen geometries required the use of different gauge lengths, in-situ monitoring of deformation/strain was also utilised in order to document local strains.In this work, the 'specimen' refers to the experimental piece of material dedicated for mechanical testing, while 'sample' refers to material subjected to metallographic analysis.

L-PBF IN-718 blade
An SLM 280 L-PBF additive manufacturing system was used to produce small turbine blades using L-PBF pedigree gas-atomised IN-718 spherical powders with an average particle size of 15-45 µm.A standard L-PBF raster strategy was used to deposit the turbine blade in the manner shown in Figure 1 with no bed preheat.The main processing parameters included 30 µm layer thickness, 0.12 mm hatch spacing, 200 W laser power and 900 mm/s scanning velocity.
Figure 1 also shows the locations of M-TT specimens subsequently excised from three specific locations (i.e.attachment, transition, blade) and two specimen orientations for each location in order to also document any anisotropy.The M-TT specimen geometry successfully used previously in other work (Džugan, Procházka, et al. 2015) is illustrated in Figure 2. In this study, machined M-TT specimens were tested in the as-built condition.

EB-PBF Ti-6Al-4V propeller
Figure 3 shows the EB-PBF-processed Ti-6Al-4V propeller, from which the M-TT specimens, Figure 2, were machined.The deposition was performed using an Arcam Q10 system operating under vacuum below 2 × 10 −3 mbar and at accelerating voltage of 60 kV.The deposition process was conducted under He at a pressure of about 4 × 10 −3 mbar, which was bled into the chamber to reduce electrostatic charging.Arcam Ti-6Al-4V powders with particle size range of 45-100 μm and layer thickness of 50 µm were used along with powder bed preheating to 740°C by using fast scanning with a defocused electron beam.The part contours were subsequently melted using a focused beam with scan speed of 4530 mm/s, current of 15 mA and voltage of 60 kV.The specimen interiors were then melted by scanning the surface at a constant line offset of 200 µm (batch 1).Another impeller was deposited (batch 2) with a modified line offset of 150 µm, while other parameters remained the same.The principal scan direction was alternated from the X-axis to the Y-axis between layers.The EB-PBF-produced propeller was also investigated in the as-built condition.

L-PBF H13 propeller
Fabrication of the simplified propeller with attached witness specimens, illustrated in Figure 4, was conducted using an SLM 280HL system equipped with one YLR fibre laser (IPG Photonics) with a nominal power rating of 400 W.During each build cycle, a 30 µm layer of the input powder (H13 tool steel of particle size: 10-45 µm; supplied by SLM Solutions AG) was spread using a dedicated recoating device.Subsequently, the powder layer was exposed to the laser beam in areas defined by the current cross-section of the model and repeated until completion.The processing parameters for H13 tool steel were 610 mm/s and 175 W for scan speed and laser power, respectively.The propeller scanning strategy used a stripe-hatch approach where the scan vectors for the core region were rotated 60°clockwise between adjacent layers.The process was conducted under nitrogen inert atmosphere and the oxygen content was held under 0.02 vol.%.The substrate base plate was pre-heated to the maximum possible temperature (200°C).
The simplified propeller was printed with attached tensile specimens in both horizontal and vertical deposition orientations in order to compare the M-TT tensile properties to those excised directly from the propeller blades.The M-TT specimen geometries used in the as-built L-PBF-H13 simplified propeller investigations are depicted in Figure 5(a and b).Straight specimens were excised from the propeller blades in order to simplify extraction while also enabling subsequent comparison to attached specimens.Although gauge length differences exist amongst the M-TT specimens, in-situ surface strain measurements were conducted to both document strain evolution and enable construction of stress-strain curves for all samples.(Džugan, Procházka, et al. 2015).

Microstructural evaluation and X-ray µcomputed tomography:
The samples for microstructural evaluation were cut from tensile specimens' shoulders with respect to the deposition planes, embedded in resin and prepared by means of standard metallographic techniques consisting of grinding and subsequent polishing.The microstructural evaluation was conducted on a JEOL 6380 scanning electron microscope (SEM) and on a Zeiss Axio Observer.Z1m optical microscope.Section cuts in XY, YZ and ZX planes were prepared by means of standard metallographic procedures.The microstructure examination of IN-718 utilised electrolytic etching in the aqueous solution of 60% HNO 3 followed by optical microscope observations on the Zeiss Axio Observer Z1m.The Ti-6Al-4V was etched with Kroll's reagent, while 3% Nital was used for H13 material.The JEOL 6380 SEM was used for analysis of M-TT fracture surfaces.
X-ray µ-CT was carried out using an YXLON FF35 system with 200 kV tube voltage and 110 µA tube current, 500 ms integration time with about 2000-2500 projections in one full rotation.A resolution of about 15-25 µm was achieved and maintained for all scans.In order to achieve this resolution, the scans were performed using a helical path strategy to scan the whole sample instead of using a multiple-section scan and stitching.Three-dimensional defect volumes were determined by rotating the sample through 360°, enabling the detector to capture numerous two-dimensional projected images from which a three-dimensional data set of volume elements (voxels) was reconstructed based on appropriate algorithms and high computing power computers (Moore 2002).Reconstruction was done at CWRU with system-supplied software including a beam hardening correction.The data set of each test sample was imported to the software application VG Studio MAX 3.0.At that point, the volume of each test sample, as well as the inner structures, was examined and each layer analysed for defects.Additional analysis options such as porosity and inclusion analysis provide  the opportunity to determine both voids and inclusions by the use of various algorithms in future work.

Micro-Tensile testing
Room temperature M-TT tests were carried out under quasi-static loading using a system with load capacity of 5 kN.The tests were run with a constant cross-head velocity of 0.1 mm/min providing an initial strain rate of 2 × 10 −4 s −1 .Strain was measured directly on each specimen with the use of a digital image correlation (DIC) system.The DIC system camera resolution was 2352 1728 pixels and was calibrated prior to testing using certified reference blocks.Local strains on the tensile specimens were measured by using a random pattern applied on the specimens' surface.The pattern on the specimens' surface was created by airbrush application of graphite paint in a stochastic pattern on white background paint.Prior to each test, the specimen dimensions were measured with the use of a micrometer.0.2% offset yield stress (OYS), ultimate tensile strength (UTS), uniform elongation at maximum force (UE), elongation (EL) and reduction of area (RA) were evaluated.Engineering stress-strain curves were constructed for each test.Specimen elongation was calculated using elongation measured from the broken  specimens' gauge length using an optical microscope.Because of this, maximum elongation displayed in the engineering stress-strain curves, where strain was measured in-situ using DIC, were often different than the elongation values shown in the tables which were calculated from the broken specimens.According to the ASTM standard, elongation measurement using two different methods is not mandatory.The elongation calculated using DIC records was included in the results as EL DIC for broader comparison of the results.DIC systems of high resolution track the material deformation with high precision.On the other hand, the manual measurement method using optical microscope can be affected by systematic errors introduced by operator lowering the precision of the method.

IN-718 blade
Figure 6 shows the microstructure details at different build locations for the as-built IN-718 blade..The microstructure exhibits remnants of the melt pools that exceed several layer thicknesses in the YZ plane and coarse columnar grains, while the largest grain width (e.g. 100 µm) is observed in the attachment region.The melt pool widths in the YZ plane exhibited a maximum of 200/150/150 µm for the attachment/transition/blade regions, respectively.Microstructural features in the XZ plane were spaced roughly 25-50 µm for the same regions.
Figure 7 and Table 1 summarise the differences in the IN-718 local blade properties with the use of M-TT specimens in both longitudinal and transverse directions.The specimens oriented along the build direction (i.e.longitudinal specimens ZYX and ZXY) yield differences of about 200 MPa in both strength-related parameters (i.e.OYS, UTS) over the three sampling locations (i.e.attachment, transition, blade).Smaller strength differences (e.g.50 MPa) were exhibited in the horizontal build orientation (i.e.transition specimens YZX) among the different sampling locations.The smallest differences in OYS/UTS strength values between the horizontal (i.e.transverse -YZX) and vertical (i.e.longitudinal -ZYX and ZXY) build orientations were observed for the blade regionspecimens 5-YZX (721/1073 MPa) and 6-ZYX (718/967 MPa), while the biggest differences

Ti-6Al-4V propeller
Figure 3 shows the as-built EB-PBF Ti-6Al-4V propeller along with locations of specimens excised from the core and each of the blade regions The M-TT specimens    taken from the core region were aligned along the build (i.e.vertical Z direction) while the blade specimens were roughly perpendicular to the core specimens.Each blade produced one tensile specimen, enabling determination of property variations around the circumference of the propeller.Two as-built turbines from two different production batches were evaluated.The first one, designated as '1', was made during the initial trials, while the second one, designated as '2' was produced using the modified line offset value described earlier.The specimen IDs used for the blades are sequentially numerical (i.e. 1, 2, 3, etc.) while the centrally-  located specimens are also marked with a 'C'.Thus, 1-1 designates blade number 1 from the initial trial, while 2-1 designates blade number 1 from the modified processing trial.
Figure 11(a-d) shows the microstructure of the asbuilt EB-PBF Ti-6Al-4V propeller (batch 1), with prior β grains widths ranging from 50 to 300 µm aligned along the build (i.e.Z) direction along with thin grain boundary alpha (α GB ).The columnar prior β grains generally contained a lamellar α+β microstructure with details dependent on sample location.In the thin blade regions that experienced faster cooling, the α GB and α-lamellae thickness are smaller with measured thickness of 0.55 ± 0.25 µm (Figure 11(c)), in comparison to the thicker and slower-cooled central regions of the propeller that exhibited coarser α GB and α-lamellae 1.75 ± 0.75 (Figure 11(d)).
Figure 12 summarises the M-TT test results (i.e. with gauge length of 2 mm) for two sampling orientations and locations.In the case of batch 1 (Figure 12(a)), there are noticeable differences between tests performed on individual blades around the circumference of the propeller as well as the central hub.In contrast, almost identical stress-strain curves (Figure 12(b)) were exhibited for the modified propeller build used for batch 2, regardless of the sampling location and orientation.The modified build (with reduced line offset value from 200 to 150 µm) typically exhibited lower values for OYS/UTS and higher elongation/RoA compared to the unmodified build (i.e.batch 1).It must also be noted that the tensile axis is along the build (i.e.Z) direction for the core 'C' specimens and roughly perpendicular to the build direction for the blade specimens for modified build parameters.Tables 2 and 3 summarise all of the results.
Fracture surfaces of the EB-PBF Ti-6Al-4V propeller specimens revealed isolated internal defects from the EB-PBF deposition process, presented in Figures 13  and 14.The fracture mechanism in all specimens was transgranular ductile/dimpled failure.The dimples in the Ti-6Al-4V specimens are much coarser (e.g.3-5 μm) in comparison to those exhibited (e.g.sub-μm) in the IN-718 blade.

H13 propeller
The as-built L-PBF-H13 propeller with a simplified geometry, shown in Figure 4, was also produced in order to provide easier extraction of M-TT specimens, enabling straightforward comparison between as-built specimens  and those excised/machined from the propeller.In this case, the as-built specimens were attached to the propeller blades (Figures 4 and 5) to explore the possibility of using them as witness specimens that could be printed along with components.The small-size attached coupons used presently do not impact the build volume in the chamber while also significantly lowering the costs associated with the production of standard-sized specimens that might neither represent the local thermal conditions nor stress states present in the component.The as-built small coupons were purposely printed in two directions, Figure 4, in order to assess mechanical anisotropy, while specimens excised from the blades in the same orientations were also tested in order to provide a direct comparison to the attached witness specimens.In addition to testing the as-built attached witness specimens, some of these as-built specimens were machined to enable direct comparison to specimens excised from the as-built blade.The asbuilt attached witness specimens tested with a machined silhouette (i.e. both sides of the specimens machined) were also subsequently ground in order to examine the effects of any surface layer influence on the properties.
The microstructure (and resulting properties, fracture surfaces) of the as-built L-PBF H13 propeller, Figure 15, is consistent with a martensitic microstructure although banded features are evident and the boundaries between remnants of individual melt pools on XZ and YZ planes are still visible.The size of the prior austenite  M-TT tests were performed on both attached and excised specimens in both orientations, as shown in Figure 4.The attached specimens were investigated with as-built surfaces with gauge length of 7 mm, Figure 5(a), while specimens with a machined silhouette had a 5 mm gauge length and those with completely machined surfaces had a 4 mm gauge length.In order to maintain specimen proportionality, the regions analysed for strain accumulation/measurement were different.Table 4 summarises the nomenclature used for specimens that were tested as-built as well as those excised and then silhouette machined vs machined completely.
Figure 16(a) shows engineering stress-strain curves for attached specimens with as-built surfaces.Smaller scatter in YS and UTS but worse properties (e.g.elongation) were exhibited for the horizontal build orientation.Once the silhouette of the attached specimens was machined, the scatter within each batch significantly decreased, Figure 16(b), while also reducing strength differences between the vertical and horizontal build orientations.When the attached specimens were subsequently machined from all sides, the horizontallydeposited specimens yielded almost identical curves, while the vertical ones exhibited very high scatter with premature fracture, Figure 16(c).
Figure 17(a) shows that excised specimens with machined silhouettes yielded very similar engineering stress-strain curves for both sampling directions with measureable differences in behaviour between horizontal and vertical build orientations.Slightly lower strength values were obtained in the vertical build orientation with lower elongation.Figure 17(b) shows that machining all sides of the excised specimens removed the significant differences in stress-strain behaviour for strains up to 2% with strength and elongation differences evident at higher strains.Higher scatter in elongation values was exhibited by specimens built in the horizontal direction.Figure 18(a) provides a comparison of the attached and excised specimens with machined silhouette and demonstrates similar stressstrain curves for both the attached and excised specimens tested in the vertical build orientation.However, differences in elongation are apparent for the horizontal specimens.Table 5 summarises all L-PBF-H13 M-TT tests results.The stress-strain response of the attached and excised specimens machined from all sides, Figure 18 (b) and Figure 19 reveal good agreement, with differences in OYS for both sampling orientations below 1% and below 3.5% for UTS.The vertical build exhibited lower ductility in comparison to the horizontal build, as found in all cases.
Figure 20 shows the SEM fracture surfaces analyses of the M-TT specimens extracted from L-PBF H13 propeller.These samples exhibited many process-induced defects along with transgranular brittle fracture that appeared to initiate from such defects with little elongation or reduction in area, consistent with the low strain to failure exhibited by the stress-strain curves.

Quantification of defects via high resolution tomography
The analysis of structural integrity of the presently investigated materials produced by EB-PBF and L-PBF reveals significant differences, with defects concentrated at several regions within each as-built part.In addition to Figure 21 summarising the defects' sizes and arrangement within the L-PBF IN718 blade, the defects were predominantly present in the attachment region, while their number gradually decreased along the Z axis to the blade region.The attachment region also exhibited the greatest size of pores ranging from 0.40 to 1.75 mm.In contrast, Figure 22 shows that the defects present in the EB-PBF-processed Ti-6Al-4V propeller were more evenly distributed over the whole volume of the deposited material in both the blades and centre regions.Within a single blade of the propeller, the highest pore concentrations were present in the bottom (i.e.start of build) and top (i.e.end of build) locations, while fewer pores were present in the middle portions of the build/part.In addition, the imaged defects in the EB-PBF Ti-6Al-4V component were somewhat smaller (e.g.0.25-0.60mm) in comparison to that exhibited in the L-PBF IN718 blade shown in Figure 22. Figure 23 illustrates the distribution of pores and defects in the L-PBF-processed H13 modified propeller, with pore sizes mainly in the range 0.40-0.80mm and the largest pores approaching 1.00-1.50mm.The propeller blades exhibited the highest density of pores, with much fewer pores of smaller size in the hub.
In general, the L-PBF-processed IN-718 blade exhibited a greater number of large defects in comparison to the EB-PBF-processed Ti-6Al-4V propeller.Although post-processing (e.g.hot isostatic pressing (HIP)) may be effective in closing/minimizing pores not connected to the surface, and can enhance elongation values and RoA, strength losses have been reported due to tempering and coarsening of microstructural features (Šafka et al. 2016).In addition, various works (Šafka et al. 2016;Goel et al. 2019;Sadeghi et al. 2019;Eklund et al. 2018;Benzing et al. 2019;Masuo et al. 2018) have shown beneficial effects of HIP on high cycle fatigue since process-induced defects (pores, keyholes, lack of fusion (LoF)) act as fatigue initiation sites (Šafka et al. 2016;Goel et al. 2019;Sadeghi et al. 2019;Eklund et al. 2018;Benzing et al. 2019;Masuo et al. 2018) and severely reduce performance.

Discussion
These preliminary results demonstrate the potential of using extracted M-TT specimens from AM-processed parts to evaluate location-and orientation-dependent properties at different locations of interest.In addition, the possibility of printing M-TT specimens alongside an idealised propeller as 'witness specimens' has been demonstrated on L-PBF-processed H13 steel and shown to provide similar mechanical data (Figure 19) to those excised from the propeller as long as similar specimen surface conditions were tested.This was particularly evident when the specimens were completely machined.Although it is recognised that most AM parts/structures will require some post-processing (e.g.heat treatment (HT) and/or HIP), the present work focused on as-built specimens to demonstrate the general concept and usefulness of M-TT specimens to determine location-and orientation-dependent properties in parts where large scale testing of ASTM standardised specimens is not possible.Recent work conducted on post-processed materials to optimise performance, as well as on miniature fracture toughness and fatigue specimens, have been reported elsewhere (Machry et al. 2016;Pehlivan et al. 2020;Sadeghi et al. 2019).Table 6 provides a comparison of the data generated in the current study using M-TT specimens with the literature data on as-built material.
The as-built L-PBF IN-718 exhibited properties similar to those reported elsewhere on as-built bulk specimens (Lewandowski and Seifi 2016;Deng et al. 2018;Seifi et al. 2018).Although AM of IN-718 typically requires postprocessing (e.g.HT and/or HIP) in order to develop a more uniform and equilibrium microstructure (Deng et al. 2018;Seifi et al. 2018) with a consistent combination of various mechanical properties (e.g.tension, toughness, fatigue, etc.), the present preliminary work was conducted to demonstrate that M-TT specimens excised from bulk parts can exhibit similar tensile properties to the bulk printed specimens, with the exception of elongation values.It is known that the ductility, as measured by elongation, is gauge length dependent and thus these values must be used with caution.It must also be pointed out that true polycrystalline behaviour requires at least 10 grains across the specimen cross-section and caution must be used on any miniature specimens if this condition is not met.However, it remains important to evaluate the mechanical properties of thin part regions despite the potential lack of sufficient grains through the cross-section since that is the relevant local material condition in that region.It should also be pointed out that this condition most strictly applies to single-phase structures where the grains do not contain any substructure constituents (typically, for example, low carbon steel with pure ferritic grains that have no substructures in the ferrite grains).The presence of substructure, particles or a multi- phase microstructure within the grains, as is the case for the materials tested in this experiment, reduces the dominant effects of testing fewer grains.For example, the IN-718 columnar grains also have a cellular-dendritic substructure with different growth directions within one grain.The prior β grains in Ti-6Al-4V contain a fine α + β lamellar structure along with grain boundary α, while the high temperature austenite grains in the H13 alloy will form a fine martensitic structure containing thin martensite laths.In addition, the orientations of the α-lamellae in Ti-6Al-4V and the martensitic laths in the H13 are also different within a single grain.
The ability of the M-TT specimens to reproduce bulk strength (OYS/UTS) behaviour while providing the possibility of quickly evaluating both location-and orientation-dependent properties provides a number of opportunities to exploit, as further demonstrated by their use on the L-PBF-processed IN-718 blade, EB-PBFprocessed Ti-6Al-4V propeller and L-PBF-processed H13 modified propeller.
The use of M-TT specimens on the as-built L-PBF-processed IN-718 blade clearly revealed both location-and orientation-dependent properties in the attachment, transition and blade regions.These anisotropies partly   relate to the different microstructural scales in addition to the differences in process-induced defects, which were concentrated mostly in the attachment region and exhibited sizes that reached a maximum of 1.75 mm.While post-processing via HT may homogenise the properties, the variations in defect density and type will not be greatly affected unless HIP is used, and the effectiveness of HIP will depend on the type of defect (keyhole, gas/ shrinkage porosity, LoF) present, as well as if they are surface-emergent defects in which case HIP will be ineffective.The effects of HT and HIP on the defects density and microstructure changes and resulting mechanical behaviour of the component is out of the scope of this study and will be performed in future works.
The presence of process-induced defects in AM-processed products is primarily dependent on the quality of powder feedstock (e.g.Ar atomised powder that may contain spherical pores) and selection of deposition parameters (e.g. that may produce LoF, keyhole porosity, etc.).In the case of deposition parameter-related issues, the number of defects in the material can be successfully reduced by optimisation of laser power, scanning velocity, hatching distance and layer thickness (Narra et al. 2022).The higher incidence of defects in thin walls of deposited components (e.g.propeller blade regions) in comparison to bulk deposits may be enhanced due to the inclination of the propeller wall and insufficient laser track overlapping.While this issue could be solved by increasing the size of melt pool via increased power input, this could move one out of the process window and potentially lead to keyhole porosity.The presence and size of pores in the limited gauge regions of such miniature specimens clearly affect the repeatability of the results (elongation and RoA) since the fracture path typically links regions of the specimens weakened by the presence of defects.This will increase scatter in the elongation values as demonstrated in the attachment region of the IN-718 blade.
The M-TT specimens excised from the as-built EB-PBF Ti-6Al-4V propeller similarly revealed differences between the blades and core region, particularly in the unmodified processing conditions.The use of powder bed preheating in the EB-PBF process provides a somewhat slower cooling rate than that obtained in the L-PBF process, which produces differences between the resulting microstructures and defect distributions within the core regions and the blades, thus also affecting their mechanical response.The EB-PBF Ti-6Al-4V propellers were deposited with two different line offset values (200 and 150 µm) that will alter both the porosity and thermal history of the build.In the unmodified build (batch 1), the much finer microstructure in the blades compared to the core region produced higher strengths with lower ductility compared to the modified build (batch 2).Batch 2 samples exhibited more uniform and slightly lower OYS/UTS properties and closer to those reported in the literature on as-built EB-PBF Ti-6Al-4V (Lewandowski and Seifi 2016; Pehlivan et al. 2020).In addition, the blades in the unmodified Ti-6Al-4V propeller exhibited a finer microstructure with lower strength in comparison to the core region, although the blades were oriented at 90 degrees to the core samples.In this case, the M-TT specimens revealed that the grain size is not necessarily the dominant parameter affecting the mechanical characteristics, but specimen orientation with respect to the build direction is also crucial.The fracture surfaces clearly revealed the presence of porosity in the EB-PBF Ti-6Al-4V (batch 1), although the bulk of the M-TT specimens failed by  dimpled/ductile fracture, consistent with previous work on EB-PBF Ti-6Al-4V that uses a pre-heated powder bed at temperatures similar to what was used presently (Lewandowski and Seifi 2016).
The demonstration of the use of the M-TT specimens for the as-built L-PBF-processed H13 modified propeller revealed the possibility of using such specimens as 'witness specimens' printed alongside parts, although the surface condition used for testing must also be considered.Although H13 steel is typically used in the quenched and tempered condition to produce a balance of strength, ductility and toughness for use in a variety of die applications, the approach used presently was conducted on as-built material to demonstrate the general approach of using M-TT specimens to quantify location-and orientation-dependent properties in addition to exploring the concept of 'witness specimens' extracted from the part attachment that can be potentially extended to other materials/part families.The goal in such work is to ensure that the  attachment has the same thermal history as the manufactured component so that investigation of the microstructure and mechanical properties of the attachment can be used to track the process quality.
The L-PBF-processed H13 clearly revealed defects and microstructural banding, the latter likely due to microsegregation.The high rates of cooling (i.e.due to lack of significant powder bed preheat) in the L-PBF process typically produce martensitic microstructures that are very strong but brittle, although the multiple passes required to produce products in the AM process likely provides some in-situ tempering.The strengths of the as-built specimens were in the range of those reported for H13, however the ductility was significantly lower due to the combinations of process-induced defects and non-optimised microstructure (i.e.banded microstructure, lack of subsequent tempering, etc.).The fracture surfaces clearly showed brittle features that emanated from the process-induced defects, consistent with the low elongation values and RoA (e.g.1%/3%) that is much less than that of conventional H13 in the optimised quenched and tempered condition.The tomography results on the H13 propeller demonstrated that numerous local defects ranging from 0.40 to 1.50 mm in size were concentrated mostly in the propeller blades.Future work will examine post-processed materials in order to begin a systematic evaluation of the factors controlling the location-and orientation-dependent property trade-offs using this material and M-TT specimens.
The results obtained within this study across all materials/parts are compared with the data published on bulk specimens by other researchers in Table 7.The differences between bulk testing and those excised from as-built parts continues to highlight the need to understand factors contributing to locationand orientation-dependent properties in actual components.

Conclusions
In the present study, three additively manufactured parts made of L-PBF-processed IN-718, EB-PBF-processed Ti-6Al-4V and L-PBF-processed H13 tool steel alloys were investigated.
In the case of L-PBF-processed IN-718 blade, a complex mechanical performance was exhibited across the part, demonstrating heterogeneous properties that were both location-and orientation-dependent.The behaviour of EB-PBF-processed Ti-6Al-4V propeller was evaluated for unmodified and modified deposition parameters.The unmodified processing conditions produced a part that exhibited higher strength values but lower elongation in comparison to that created with a modified process to create the same propeller.The specimen orientation appeared to be the main influencing parameter since the specimens from the core region were excised longitudinally to the build direction.The specimens extracted from L-PBF-processed H13 propeller in the as-built state exhibited the poorest mechanical performance, nevertheless, machining reduced the degree of mechanical properties scatter.
The concept of attaching 'witness specimens' utilised presently provides a potential low-cost AM process/part quality check.The stress-strain response of the attached and excised specimens machined from all sides produced very good agreement of the mechanical behaviour.Comparison of strength values for the attached and excised specimens with machined silhouette and completely machined ones produced a maximum difference of about 5% (for one data sethorizontal direction, machined silhouette) and for the other three data sets it was below 2.5%.These results confirm the potential of using such attached specimens utilisation for quality check of AM-processed parts.
In conclusion, this preliminary work was conducted to first demonstrate the usefulness of conducting M-TT tests on excised specimens from parts to determine location-and orientation-dependent properties.In addition, the concept of using miniature specimens attached to a part as a witness specimen was proposed.If the thermal conditions in the 'witness specimens' can be processed to match those of the component, this could provide reliable information on the properties of the component without destructive testing of the component.In the present work, this yielded similar results when M-TT testing was conducted with similar surfaces.

Figure 1 .
Figure 1.Sampling scheme for M-TT specimens excised from IN-718 blade.Z-build orientation.

Figure 2 .
Figure 2. Micro-tensile test specimen geometry used presently and successfully used in the previous work (Džugan, Procházka, et al. 2015).

Figure 5 .
Figure 5. M-TT specimen geometries used in the as-built L-PBF H13 simplified propeller investigations: (a) dimensions of as-built specimens that were attached to the propeller and (b) specimens excised from the propeller.Dimensions in mm.

Figure 6 .
Figure 6.Microstructure of as-built IN-718 blade from (a) attachment region, (b) transition region, (c) blade region and (d) detail of microstructure in the YZ plane of the blade region (i.e.image (c)).

Figure 7 .
Figure 7. Engineering stress-strain curves for M-TT specimens at various locations and orientations excised from IN-718 blade.

Figure 9 .
Figure 9. Fracture surface of M-TT specimen extracted from the IN-718 transition region (specimen 3-YZX): (a) overview of the whole fracture surface; (b) detail of an internal lack of fusion defectcavity and (c) locally ductile appearance of fracture surface exhibiting very fine dimples.

Figure 8 .
Figure 8. Fracture surface of M-TT specimen extracted from the IN-718 blade region (specimen 5-YZX): (a) overview of the whole fracture surface; (b) detail of an internal defectcavity and (c) locally ductile appearance of fracture surface illustrating very fine dimples.

Figure 11 .
Figure 11.Microstructure of the EB-PBF Ti-6Al-4V propeller (batch 1): (a) blade region; (b) centre region; (c) detail of YZ plane in blade region and (d) detail of YZ plane in the centre region.

Figure 10 .
Figure 10.Fracture surface of M-TT specimen extracted from the IN-718 blade attachment region (specimen 2 ZYX): (a) overview of the whole fracture surface; (b) detail of an internal defectcavity and (c) locally ductile appearance of fracture surface exhibiting very fine dimples.

Figure 12 .
Figure 12.Comparison of M-TT stress-strain curves for as-built EB-PBF Ti-6Al-4V propeller blades and central region -(a) batch 1 for unmodified build parameters and (b) batch 2 for modified build parameters.

Figure 14 .
Figure 14.Fracture surface of as-built EB-PBF Ti-6Al-4V M-TT specimen propellercentre (1C_3): (a) overview of the whole fracture surface; (b) detail of an internal defectcavity and (c) locally ductile appearance of fracture surface.

Figure 15 .
Figure 15.Microstructure of the as-built L-PBF-H13 propeller blade region: (a) 3D overview of the microstructure and (b) detail of internal LoF defects.

Figure 16 .
Figure 16.Engineering stress-strain curves for as-built L-PBF-H13 propeller attached specimens with (a) as-built surfaces, (b) machined silhouette and (c) all sides machined.

Figure 16
Figure 16 Continued

Figure 17 .
Figure 17.Engineering stress-strain curves for as-built L-PBF-H13 propeller -(a) excised specimens with machined silhouette and (b) excised specimens with all sides machined.

Figure 18 .
Figure 18.Comparison of engineering stress-strain curves for as-built L-PBF-H13 propeller -(a) attached and excised specimens with machined silhouette and (b) attached and excised specimens with completely machined surfaces.

Figure 19 .
Figure 19.M-TT test results for L-PBF H13 propeller: (a) strength values of the specimens produced in horizontal orientation; (b) elongation and reduction of area values of the specimens produced in horizontal orientation; (c) strength values of the specimens produced in vertical orientation; (d) elongation and reduction of area values of the specimens produced in vertical orientation and (e) legendbatch designation system.

Figure 20 .
Figure 20.Fracture surface of as-built L-PBF-H13 propeller -M-TT vertical specimen ZM6: (a) whole fracture surface and (b) detail of internal defects present at the fracture initiation site.

Figure 21 .
Figure 21.X-ray µ-CT resultsdefect sizes and distribution of as-built L-PBF IN-718 blade.Higher pore density evident in the attachment region.

Figure 22 .
Figure 22.X-ray µ-CT resultsdefect sizes and distribution of as-built EB-PBF Ti-6Al-4V propeller.Higher pore density exhibited at the start (i.e.bottom) and top (i.e.end) of the build/part.

Figure 23 .
Figure 23.X-ray µ-CT resultsdefect sizes and distribution for as-built L-PBF H13 modified propeller.Higher pore density exhibited in the blades with fewer and smaller pores in the core/hub.

Table 1 .
M-TT test results for IN-718 blade.

Table 5 .
M-TT test results for L-PBF H13 propeller.

Table 6 .
Comparison of data from current study with literature data for as-built state.

Table 7 .
Comparison of obtained results with published data.