SE1950517A1 - Process for producing a steel workpiece by additive powder bed fusion manufacturing, and steel workpiece obtained therefrom - Google Patents

Abstract

The present disclosure relates to a process for producing a steel workpiece having an elongation at fracture of greater than 70%, as determined by the method of ASTM Test E8-16a. The process comprises the steps:- providing a steel powder comprising an active metal; and- subjecting the steel powder to an additive powder bed fusion manufacturing process, thereby producing the steel workpiece.The disclosure further relates to use of a steel powder comprising an active metal in such a process, as well as a steel workpiece obtained by such a process.

Description

1Process for producing a steel workpiece by additive powder bed fusion manufacturing, and steel workpiece obtained therefrom TECHNICAL FIELD The present disclosure relates to a process for producing a steel workpiece by additive powderbed fusion manufacturing. The disclosure further relates to use of a steel powder comprising an active metal in such a process, as well as a steel workpiece obtained by such a process.

BACKGROUND ART Metal additive manufacturing, colloquially termed 3D printing, is capable of being used tofabricate a variety of components having structures of such complexity that they are unviableor impossible to produce by traditional casting or subtractive manufacturing (machining)methods. Metal additive manufacturing therefore has the potential to find utility in a large range of fields including medical implants, engines and automobile components.

Selective laser melting (SLM) is an additive manufacturing process utilizing a focused laserbeam to melt and fuse metal powder together in a localised melt pool. Due to the metalpowder being fully melted, solid metal components are formed in the process, in contrast toadditive sintering methods such as selective laser sintering. To date, the metals that have beendemonstrated in SLM processes include steel, cobalt-chrome, inconel, aluminium and titanium powde rs.

A known drawback of SLM processes is however that components manufactured by thetechnique tend to have inconsistent and unreliable mechanical behaviour. This limitation hinders the more widespread implementation of SLM as a manufacturing technique.

There remains a need for a means for metal additive manufacturing that allows for the production of components with reliably excellent mechanical properties.

SUMMARY OF THE INVENTION The inventors ofthe present invention have identified shortcomings with prior art means ofmetal additive manufacturing. The inventors have identified that the inconsistent andunreliable mechanical behaviour of components manufactured by SLM is due at least partly tothe presence of residual oxygen in the additive manufacturing apparatus during the additivemanufacturing process. Although metal additive manufacturing is typically carried out underinert conditions or under vacuum (depending on the additive manufacturing method utilized),it is all but impossible to avoid the presence of at least some residual oxygen both from theprocessing chamber and the precursor powder. This residual oxygen may manifest as an oxidelayer at the grain boundaries ofthe produced component, leading to weakened grain boundaries and degraded mechanical properties. lt is an object of the present invention to achieve a means of overcoming, or at leastalleviating, the above-mentioned shortcomings. ln particular, it is desired to provide a meansof additive manufacturing that avoids or reduces the formation of oxides at the grain boundaries of a manufactured component.

These objects are achieved by a process for producing a steel workpiece according to the appended independent claim.

The steel workpiece produced by the process has an elongation at fracture ofgreater than 80% as determined by the method of ASTM Test E8-16a, i.e. it has exceptional ductility.

The process comprises the steps of providing a steel powder comprising an active metal; andsubjecting the steel powder to an additive powder bed fusion manufacturing process, thereby producing the steel workpiece.

By steel workpiece it is meant any finished or unfinished component producible by metaladditive manufacturing of a steel powder. For example, in order to produce a finishedcomponent from the workpiece, one or more further steps such as machining or coating may be required.

The inventors have discovered that the provision of an active metal in the steel powder consistently leads to the produced workpieces having exceptional ductile properties as 3defined above. I\/|oreover, the produced workpieces also have consistently excellent strength, both yield strength and tensile strength.

Without wishing to be bound by theory, it is thought that the presence of an active metal inthe steel powder allows the active metal to be sacrificially oxidized by residual oxygen duringthe additive manufacturing process. The resulting active metal oxide particles are dispersed asnanoparticles throughout the metal, accompanied with the consumption of oxygencontamination inside the grain boundaries and the elimination of large oxide precipitation thatotherwise may form at the grain boundaries .This means that the grain boundaries aresubstantially free of oxide and are strengthened in comparison to oxide-coated boundaries.|nstead of weakening the produced component, it is thought that the dispersed active metaloxide strengthens the metal component, and the resulting workpiece can thus be considered to be formed of an oxide-dispersion strengthened (ODS) alloy.

By an active metal it is meant any metal where the sacrificial reaction of oxygen with theactive metal is favoured over the reaction of oxygen with the iron in the steel powder at the concentrations and conditions prevailing during the additive manufacturing process.

According to a further aspect, the objects of the invention are achieved by the use of a steelpower as defined in the appended independent claim. The steel powder comprises an activemetal, and is used in an additive powder bed fusion manufacturing process for producing asteel workpiece having an elongation at fracture of greater than 80%, as determined by the method of ASTM Test E8-16a.

According to yet a further aspect, the objects of the invention are achieved by a steelworkpiece as defined in the appended independent claim. The steel workpiece is produced by the process as defined herein.

These further aspects have the same advantages as described in relation to the inventiveprocess as described herein, i.e. improved reliability, exceptional ductility and excellent strength ofthe produced steel workpieces.

The following features are applicable to the process, the use and the steel workpiece as defined in the appended independent claims, unless otherwise stated. 4The steel workpiece having an elongation at fracture of greater than 80%, as determined bythe method of ASTM Test E8-16a, may have an eleongation of at least 85%, or in some cases an elongation of at least 90%.

The additive powder bed fusion manufacturing process may be selective laser melting,electron beam melting, selective laser sintering, selective heat sintering, or direct metal lasersintering. Processes whereby the metal powder is locally melted during processing, such asselective laser melting or electron beam melting, are preferred. Selective laser melting is especially preferred.

The steel powder may be any appropriate steel powder, such as an austenitic steel powder.

Austenitic steel powders ofthe 300 series, such as 316L steel powders, are preferred.

The active metal may be constituted of a metal alloyed in the steel powder. This metal may be selected from Si, Cr, Mo, Ti, Ta and Nb. Silicon is preferred.

The active metal may be added in a quantity sufficient to react substantially completely withany residual oxygen present during the additive powder bed fusion manufacturing process.This assists in ensuring grain boundaries that are substantially free of large oxide particles in the workpiece and thus assists in improving the mechanical properties of the workpiece.

The steel powder may comprise a nanoparticle metal oxide. The nanoparticle metal oxide maypreferably be yttrium oxide nanoparticles. By adding a metal oxide nanoparticle secondaryphase to the steel powder, a stronger oxide-dispersion strengthened (ODS) alloy is obtained in the resulting workpiece, and the mechanical properties ofthe workpiece are enhanced.

The steel workpiece may have a yield strength of at least 500 MPa, such as at least 550 MPa,as determined by the method of ASTM Test E8-16a. Such yield strengths are in excess ofthose typically obtained in steel workpieces from additive manufacturing. lt is well establishedthat there is typically a trade-off between strength and ductility, and steel workpieces combining both exceptional ductility and excellent strength are highly unusual.

The steel workpiece may have a tensile strength of at least 580 MPa, such as at least 590 MPaor at least 600 MPa, as determined by the method of ASTM Test E8-16a. 5The additive powder bed fusion manufacturing process may comprise a step of rotating ascanning direction between adjacent layers. For example, the scanning direction of each layermay be offset from the immediately preceding layer by from about 45° to about 90°, such asabout 67°. The microstructure and mechanical properties ofthe workpiece may be customized by offsetting the scanning direction in each layer in such a manner.

Further objects, advantages and novel features of the present invention will become apparent to one skilled in the art from the following detailed description.

BRIEF DESCRIPTION OF THE DRAWINGS For a fuller understanding of the present invention and further objects and advantages of it,the detailed description set out below should be read together with the accompanyingdrawings, in which the same reference notations denote similar items in the various diagrams, and in which: Figure la schematically illustrates the layout of the specimens showing the dimension and laser scanning strategy.Figure lb is a photograph illustrating the as-prepared tensile test specimens.

Figure 1c is a photograph illustrating the machined specimens with gouge length of 10 mm and gauge cross-section of 1><1 mm for tensile tests.Figure 1d is a photograph illustrating the tensile test facility for small sized specimens.

Figure 2a is a graph illustrating the cell spacing, density and Vickers hardness ofthe various small-scale specimens.

Figure 2b is SEM images of the etched surfaces showing the cell spacing decreases as scanning speed increases.Figure 3a shows a SEM image of PH45 at a first magnification.Figure 3b shows a SEM image of PH90 at a first magnification.

Figure 3c shows a SEM image of PH45 at a second magnification. 6 Figure 3d shows a SEM image of PH90 at a second magnification.

Figure 3e shows a figurative illustration of the cell growth in the various layers of PH45.Figure 3f shows a figurative illustration ofthe cell growth in the various layers of PH90.Figure 4a shows an EBSD orientation map of PH45 in a first direction.

Figure 4b shows an EBSD orientation map of PH45 in a second direction.

Figure 4c shows the inverse pole figures of the top surface of PH45.

Figure 4d shows an EBSD orientation map of PH90 in a first direction.

Figure 4e shows an EBSD orientation map of PH90 in a second direction.

Figure 4f shows the inverse pole figures of the top surface of PH90.

Figure 5 shows the tensile engineering stress-strain curve of SLM specimens PH90 (blue), PH45 (red) and PV45 (green).

Figure 6a figuratively illustrates the fracture tip of PH45, with an OM image inserted in the figure.

Figure 6b is a SEM image illustrating the fracture tip of PH45 in a side view of fracture tip showing cell direction and melt pool boundaries.

Figure 6c is a SEM image illustrating the fracture tip of PH45 in a top view showing different fracture mode at rupture.

Figure 6d is a SEM image illustrating the fracture tip of PH45 in a magnified images showing details.

Figure 6e figuratively illustrates the fracture tip of PH90, with an OM image inserted in the figure.

Figure 6f is a SEM image illustrating the fracture tip of PH90 in a side view of fracture tip showing cell direction and melt pool boundaries. 7Figure 6g is a SEM image illustrating the fracture tip of PH90 in a top view showing different fracture mode at rupture.

Figure 6h is a SEM image illustrating the fracture tip of PH90 in a magnified images showing details.

Figure 7 shows the derived true stress-strain curves and the strain hardening curves of different specimens (PH90, PH45 and PV45).Figure 8a is an EBSD image revealing the grain boundaries and the sub-grain boundaries.

Figure 8b is a SEM image ofthe etched surface at the same site showing the ce||u|ar structure and the melt pools.

Figure 8c is a magnified SEM image of site c as marked in Figure 8a.Figure 8d is a magnified SEM image of site d as marked in Figure 8a.Figure 8e is a magnified SEM image of site e as marked in Figure 8a.

Figure 8f is a schematic drawing of the hierarchical structures in AM materials.

DETAILED DESCRIPTION Additive powder bed fusion manufacturing processes involve the layer-by-layer build-up of aworkpiece by depositing a layer of metal powder, such as steel, and then precision meltingselected areas ofthe powder layer using a heat source. For example, in selective laser melting(SLM) processes the heat source is a laser, typically focused by a lens and precision-directedusing mirrors. A further layer of powder is then deposited and the melting step repeated inorder to provide the next layer of metal fused upon the original layer. The deposition andmelting steps are then alternatingly repeated until the desired metal workpiece has beencompleted. The workpiece is removed from the surrounding bed of non-fused powder and may then be subjected to further treatment in order to provide a finished component.

The processing parameters such as laser power, scanning speed, hatch spacing and layer thickness are known to affect the mechanical properties ofthe produced workpiece, and 8 these parameters should be tuned for each powder in order to provide the desired optimizedmechanical properties. For example, the laser power may be from about 150 W to about 300W, preferably about 200W. The scanning speed may be from about 500 mm/s to about 7000mm/s, preferably from about 700 to about 900 mm/s. The hatch spacing (line separationdistance) may be from about 0.01 mm to about 0.15, preferably from about 0.06 mm to about0.15 mm. The layer thickness may be from about 0.01 mm to about 0.03 mm, preferably about 0.02 mm.

The microstructure and mechanical properties ofthe workpiece may also be customized bychanging the scanning direction in each layer. For example, the scanning direction of eachlayer may be offset from the immediately preceding layer by from about 45° to about 90°, such as preferably about 67°.

The present invention is based upon the discovery by the inventors that providing an activemetal in the steel powder allows workpieces to be produced that have exceptional ductilityand excellent strength. By an active metal it is meant any metal where a sacrificial reactionof oxygen with the active metal is favoured over the reaction of oxygen with the iron in thesteel powder at the concentrations and conditions prevailing during the additivemanufacturing process. Suitable active metals include, but are not limited to alloying metals such as Si, Cr, I\/|o, Ti, Ta and Nb. One or more of these metals may be present as an alloy metal in the steel powder.

Although metal additive manufacturing is typically performed under inert conditions (inertatmosphere or vacuum), there is typically some residual oxygen present during the additivemanufacturing process, for example in the inert atmosphere (which may e.g. have up to 10000ppm oxygen) or associated with the steel powder. lt is thought that this residual oxygen reactswith the steel powder during the additive manufacturing process, forming an oxide layer atthe metal grain boundaries and thus weakening the produced workpiece. Without wishing tobe bound by theory, it is thought that the presence of an active metal in the steel powderallows the active metal to be sacrificially oxidized by residual oxygen during the additivemanufacturing process. The resulting active metal oxide is dispersed as a nanoparticlethroughout the metal and is not localised at the grain boundaries. This means that the grain boundaries are substantially free of oxide when a steel powder comprising an active metal is 9used, and the grain boundaries are therefore strengthened in comparison to oxide-coatedboundaries. lnstead of weakening the produced component, it is thought that the dispersedactive metal oxide somewhat strengthens the metal component, and the resulting workpiece can thus be considered to be formed of an oxide-dispersion strengthened (ODS) alloy.

The active metal may be added as a secondary phase to the steel powder, or alloyed in thesteel powder. lt is preferably added or alloyed in quantities sufficient to react withsubstantially all residual oxygen in the additive manufacturing process. This may bedetermined by the skilled person by using steel powders having a variety of concentrations ofactive metal and determining which powder provides workpieces having optimal mechanicalproperties. Alternatively, a theoretical quantity of active metal required may be determinedfrom the stoichiometry ofthe reaction of the active metal with oxygen, provided that theamount of residual oxygen is known. The steel powder may for example comprise from about 0.1% to about 0.9% by weight of one or more active metals in total.

The invention will now be described in more detail with reference to certain exemplifyingembodiments and the drawings. However, the invention is not limited to the exemplifyingembodiments discussed herein and/or shown in the drawings, but may be varied within thescope ofthe appended claims. Furthermore, the drawings shall not be considered drawn to scale as some features may be exaggerated in order to more clearly illustrate certain features.

ExamplesExample 1: Small-scale specimens (40 mm X 4 mm) All specimens were prepared using a gas atomized spherical SS316L powder with particle sizeranging from 10 to 45um (Carpenter powder products AB, Torshälla, Sweden) as the precursor powder. The powder alloy comprises approximately 1% by weight silicon as the active metal.

The specimens were prepared by a commercial SLM system EOS I\/|270 (EOS GmbH, Krailling,Germany). The experimental details ofthe various specimens are shown in Figs. 1a-d with the processing parameters listed in Table 1.

Figure 1a schematically illustrates the layout of the specimens showing the dimension andlaser scanning strategy, where BD represents building direction and TD represents transversedirection. Figure 1b is a photograph i||ustrating the as-prepared tensile test specimens. Figure1c is a photograph i||ustrating the machined specimens with gouge length of 10 mm and gaugecross-section of 1><1 mm for tensile tests. Figure 1d is a photograph i||ustrating the tensile test facility for small sized specimens.

Table 1. Processing parameters of the test specimens and the calculated relative density (theoretical density of S531 6L is 8000 kg/m3) Specimen Laser Scanning Hatch Layer Energy Rotation RelativeNo. power speed (v) spacing thickness density angle (°) density(W) (mm/S) (dllmm) (hllmm) (J/mma) (%) C1 195 7000 0.01 0.02 139 45 95.3 CZ 195 4250 0.02 0.02 114.5 45 98.6C3 195 1700 0.05 0.02 114.5 45 99 C4 195 850 0.1 0.02 114.5 45 99.2 C5 195 566 0.15 0.02 114.5 45 99.5PH45 195 850 0.1 0.02 114.5 45 99.2PH90 195 850 0.1 0.02 114.5 90 99.3PV45 195 850 0.1 0.02 114.5 45 99 The laser melt traces were characterized on the etched surface by Light Optical I\/licroscopy(LOM). Etched surfaces of both as-prepared specimens and the specimens after tensile teststogether with the fracture surface were observed by JEOLJSM-7000F field emission scanningelectron microscopy (SEM) (JEOL, Tokyo, Japan). Prior to microstructure observation, thespecimens were mounted, polished and etched in 50 ml HF solution (HFzHNOgzHzO = 2:6:42)for 5 mins. EBSD was performed on a HKL Nordlys orientation imaging microscope system(Oxford Instruments, Oxford, UK) equipped on a TESCAN I\/IIRA 3LMH with a step size less than 1.5 um.

The average cell spacing as illustrated in Fig. 2 was calculated by randomly choosing 10 sites with more than 1000 cells counted in each etched specimen. The densities of the specimens 11 were checked by Archimedes method. The as-prepared specimens were machined to dog-bone shaped tensile test specimens as seen in Fig. 1c. The gouge length was 10 mm and thecross-section size of tested part was around 1><1 mm. An extensometer was used to measurethe elongation during tests as seen in Fig. 1d. The tensile test directions with reference to as-prepared specimens were also indicated in Fig. 1a, where BD represents building direction andTD represents transverse direction. The reported values in this study for tensile propertieswere average value of 3 tests. Vickers Hardness tests were carried out at RT using aZwick/Roell ZHV indenter (Zwick/Roenhjll, Ulm, Germany) with a dual time of 10s. Five testswere performed for each specimen. A high load of 10 kgf was used to test the overall hardness and a low load of 100 gf was used to determine the local hardness.

According to solidification theory, AG/R (where AG is the temperature gradient and R is thesolidification rate) determines the solidification mode (planar, dendritic or cellular) and as wellas the dendritic arm spacing or the cell spacing. Both AG and R are related to the cooling rate,which can be tuned by the scanning speed (v) in SLM process. ln this work, we firstly tuned thelaser scanning speed and line spacing (d) to control the cell spacing of the sub-grain cellularstructure. The energy density (w=P/(vdh), P is the laser power and h is the layer thickness) was kept at a roughly same level for all the samples.

Figures 2a and 2b illustrate the effect of the laser scanning speed on the cell spacing andconsequently on the mechanical properties. Figure 2a is a graph illustrating the cell spacing,density and Vickers hardness of the various small-scale specimens. Figure 2b is SEM images of the etched surfaces showing the cell spacing decreases as scanning speed increases. lt has already been proved that the dislocation network appears together with the elementsegregation at the cell boundaries in our recent study. Therefore, the dislocation networkcharacterization which needs great amount of TEM work can be replaced by observingelement segregated cell boundaries on the polished & etched surface in SEM. Figure 2a showsthe variation of cell spacing and the bulk density. The cell spacing decreased significantly asthe increase of the laser scanning speed. I\/|eanwhile, the bulk density of the specimens wasalso influenced by the different combination of v and d. Almost full density (with a relative density of 99.2%) without obvious macro defects was achieved in specimen C4. 12 Hardness test results were carried out with a relatively high load of 10kgf in order to avoid theinfluence of the local structure. The hardness increased with increasing density until it reaches232(HV10) for sample C4 and then it dropped in C5 with even higher density (99.5%). Thedensity was still the most critical issue of concern for many applications. But when the densityvariation becomes a negligible factor, for example in C4 and C5, the balance between a higherdensity and a smaller cell spacing should be considered. ln other words, the scanning speedshould be increased when defects are already controlled at a low level, which leads to a reduced cell spacing, higher hardness and probably higher strength.

The laser scanning strategy affects the arrangement of the cellular structure. One of theimportant parameters is the rotation of laser scanning direction between successive layers. lnorder to investigate this property, we produced two batches of specimens with different rotation angles of 45° (PH45) and 90° (PH90).

Figures 3a-3f illustrate the effect of the laser scanning strategy on the morphology and thearrangement ofthe cells. Figure 3a shows a SEM image of PH45 at a first magnification. Figure3b shows a SEM image of PH90 at a first magnification. Figure 3c shows a SEM image of PH45at a second magnification. Figure 3d shows a SEM image of PH90 at a second magnification.Figure 3e shows a figurative illustration ofthe cell growth in the various layers of PH45. Figure 3f shows a figurative illustration of the cell growth in the various layers of PH90.

No obvious difference on density or macro defect was observed in both specimens. Howeverthe length and the arrangement of the cells differed significantly. The cells were noticedfrequently crossing the melt pool boundaries in PH45 (Fig. 3a, c), which resulted in'continuous' longer cells. ln contrast, most cells stopped at the melting pool boundaries andthe trend of forming longer cells was hindered in PH90 (Figure 3b, d). The growth direction ofthe cellular structure is known to incline to the local AG and may also be influenced by theI\/|arangoni convection in the melt pool [22, 23] and recent modeling work proved heat flowdirection determines the solidification texture of sub-grain structure [18]. Following the graingrowth, the cells have the chance to grow epitaxially at the melt pool boundaries and thenchange the mode to competitive growth away from the boundaries. The cells were more likelyto form longer columnar if epitaxial growth is triggered, as illustrated in Fig. 3e. When the laser rotated dramatically (900), the driving force for cells to incline to the scanning direction 13overcomes the tendency for epitaxial growth, which results in the non-continuous short cellsin Fig. 3f. This difference of cell arrangement could presumably cause significant difference in mechanical properties.

Grain structure is also an important factor of consideration. EBSD analysis (Fig. 4a-f) was doneon the two specimens. Figure 4a shows an EBSD orientation map of PH45 in a first direction.Figure 4b shows an EBSD orientation map of PH45 in a second direction. Figure 4c shows theinverse pole figures of the top surface of PH45. Figure 4d shows an EBSD orientation map ofPH90 in a first direction. Figure 4e shows an EBSD orientation map of PH90 in a second direction. Figure 4f shows the inverse pole figures ofthe top surface of PH90.

Both specimens consisted ofthe columnar grains growing in the building direction. PH90showed an ordered 'mosaic-like' pattern trapped in adjacent melt tracks (Fig. 4b) while themelt tracks were difficult to be distinguished in PH45 (Fig. 4a). While both specimens showedtexture with preferred crystallographic orientation (101) along the building direction, PH45 showed slightly stronger texture as indicated by the IPF mapping in Figure 4c, f.

Figure 5 shows the tensile engineering stress-strain curve of SLM specimens PH90 (blue), PH45 (red) and PV45 (green).

PH45 had slightly higher strength but a relative lower ductility compared with PH90. Theaverage yield strength of both specimens was high due to the hierarchical structure. Thetensile strength and yield strength for PH45 were 612 MPa and 474 MPa while those of PH90were 555 MPa and 434 MPa, respectively. The average elongations at rupture (fracture) forPH45 and PH90 were 30% and 34%, respectively, which were lower than that ofthecounterpart fabricated by traditional methods. PH45 showed better tensile strength thanPH90, which attributed to several microstructure features including grain morphology, textureand the dislocation network. The former two factors are often discussed in metals while thelast one, dislocation network, does not exist in metals from most ofthe manufacture processes, therefore rarely discussed.

The difference of the cell arrangement resulted in different fracture modes under tensile.

Figures 6a-h illustrate the performance ofthe PH45 and PH9 samples during tensile testing.

Figure 6a figuratively illustrates the fracture tip of PH45, with an OM image inserted in the 14 figure. Figure 6b is a SEM image illustrating the fracture tip of PH45 in a side view of fracturetip showing cell direction and melt pool boundaries. Figure 6c is a SEM image illustrating thefracture tip of PH45 in a top view showing different fracture mode at rupture. Figure 6d is aSEM image illustrating the fracture tip of PH45 in a magnified images showing details. Figure6e figuratively i||ustrates the fracture tip of PH90, with an OM image inserted in the figure.Figure 6f is a SEM image illustrating the fracture tip of PH90 in a side view of fracture tipshowing cell direction and melt pool boundaries. Figure 6g is a SEM image illustrating thefracture tip of PH90 in a top view showing different fracture mode at rupture. Figure 6h is aSEM image illustrating the fracture tip of PH90 in a magnified images showing details. ln thefigures, the red lines refer to the melt pool boundaries and the yellow arrows indicate the growth direction ofthe cells.

A uniform deformation of cells across the melt pools were identified in PH45 (Fig. 6b) whiledistinct shorter cells were found in PH90 (Fig. 6f). The uniform deformation featured adelamination fracture (arrow in Fig. 6c) and the fracture of long cells (Fig. 6d) was directlyobserved. One should notice the frequently formed large crater on the fracture surface wasabsent indicating PH45 had high defect tolerance. By comparison, the delamination featuredisappeared in PH90 and many large craters presented on the fracture surface (Fig. 6g). Largerstress was needed to tear apart cells as the dislocations were well pinned by the cellboundaries. Dislocation glided easier in PH90 when relative fewer dislocation walls wereblocking their movements. Therefore, a most significant strengthening effect and a higherstrength were expected in PH45. Similar phenomenon at the scale of grain level has long beenknown: more dislocation walls are encountered and a better strength is obtained when columnar grains are tensile tested perpendicular to the long columnar axis. [24, 25] Specimen PV45 was prepared by the same process parameters and strategies as PH45 butwith length of the specimen standing in building direction. The yield strength is 442 I\/|Pa andthe tensile strength is 547 I\/|Pa, both values were lower than those of specimen PH45.However, the average elongation at rupture (fracture) of PV45 reached 87% (with onespecimen reaching more than 100%), which was much longer than that of PH45. Thiscombination of strength and elongation of SLM SS316L was better than most of 316L fabricated by various methods. [26, 27] The obtained strength was attributed to the presence of dislocation network structure. ln-situTEM observation ofthe compression tests proved the cell boundaries delayed the dislocationmovement during deformation. [28] Removing cell structure by annealing leaded to adramatic decrease in yield strength. [25, 29-31] The yield strength is related to cell spacinginstead ofthe grain size following a Hall-Petch like relation. [12] Therefore, the tremendousdislocation cell boundaries played a more important role than the much fewer grainboundaries in determination ofthe strength. Similar relation as Hall-petch law between the local hardness and the cell spacing was found.

Many factors influence the ductility including grains, texture and also cellular structure. Thegrain morphology and size affects the free-path length of the dislocations during deformation.The grains and their texture also influence the deformation twinning generation. Thetremendous pre-existing dislocations at the cell boundaries limit the capacity of strainhardening but meanwhile enable stable plastic flow. lt is difficult to identify which factor takesthe leading role in determining the ductility of SLM SS316L from the present experiments.

However, we can get a clue from the dramatic ductility difference between PH45 and PV45.

Figure 7 shows the derived true stress-strain curves and the strain hardening curves ofdifferent specimens (PH90, PH45 and PV45). The strain hardening rate curves and the truestress-strain curves were derived to help understanding the ductility variation. PV45 has anobvious strain hardening rate recovery region. Twinning occur primary in grains close to <111>parallel to the tensile axis according to Schmid's law. But no preferred orientation <111>//tensile axis was observed in both samples according to the IPF mapping results (Fig. 4).Relative more grain boundaries along tensile axis in PH45 than PV45 increased the twinningstress and partially suppressed the twinning formation. [32, 33] lt concluded that the grain sizeinfluenced the twinning generation and further the strain hardening capacity. ln addition, thecell boundaries coupled with bundles of dislocations served as the nucleation sites fortwinning. The cell boundaries also stabilized the dislocation network until large strain butcouldn't fully block the dislocation motion at high stress levels. This resulted in a stable plasticflow and delayed the onset of necking. However, it was difficult to quantify the anisotropylevel of cellular structure in different tensile directions. On the other hand, the grain boundarywas known to be able to fully block the dislocation motion and deteriorate the ductility. [34] The columnar shaped grains generated much more high-angle boundaries perpendicular to 16building direction (PH45) than along building direction (PV45). The strain hardening capacitywas thus lowered and ductility dropped dramatically in PH45. The present result proved thatthe grain size and the dislocation network both influenced the ductility, although further experiments are needed to clarify the leading factor.

Figures 8a-f illustrate relations between the cellular structure, the grains and the melt pools.Figure 8a is an EBSD image revealing the grain boundaries and the sub-grain boundaries.Figure 8b is a SEM image ofthe etched surface at the same site showing the cellular structureand the melt pools. Figure 8c is a magnified SEM image of site c as marked in Figure 8a. Figure8d is a magnified SEM image of site d as marked in Figure 8a. Figure 8e is a magnified SEMimage of site e as marked in Figure 8a. Figure 8f is a schematic drawing ofthe hierarchical structures in AM materials: Melt pool boundaries (blue dashed line), High angle grain boundaries (red line), low angle grain boundary (green line) and cell boundaries (black line).

The relation between hierarchical structures has not been investigated before. The cells wereformed due to cellular grain growth under high AG/R value combined with elementsegregation at the solidification front. Therefore, the grain growth and the convection in themelt pools both influenced the arrangement of the cells. Here we introduced a new methodby comparing the grain boundaries (shown by EBSD mapping) and the cell boundariestogether with the melt pool boundaries (revealed by SEM on etched surface) at the same site(Fig. 8a, 8b). Different cases were marked in Figs. 8a-fand also summarized in Table 2. lngeneral, cells were always similar in the same sub-grain (with low-angle grain boundaries)without crossing melt pool boundary but might be different in other cases. The cells weresimilar in adjacent sub-grains (case 3) due to the sub-grains texture caused by large AG at localsite, in different melt pools (case 4) because ofthe epitaxial cell growth at melt poolboundaries. On the other hand, the cells more likely changed their arrangement if non-epitaxial grain growth occurred at the melt pool boundaries (case 6), which corresponds to thesituation in PH90. ln a word, any process parameter that changed the AG or the melt pool features influenced the cell formation and further the mechanical behavior.

Melt pool boundaries formed by layer by layer process are more like to accumulate defects.[35] Scanning strategy should enable just enough overlapping of melt pools to minimize the amount of melt pool boundaries. Grain boundary is believed to have great influence on 17strength and ductility in traditional fabricated materials. The influence of grain boundary onductility is still active but the tremendous cell boundaries vague its impact on strength. Thecell boundaries benefit both strength and ductility at certain tensile axis. Careful control of the cellular structure makes it possible to fabricate customized materials by SLM.

Table 2. The cell continuity regarding to sub-grains and melt pools (cases marked in Figs. 8a-f) Same sub-grain Different sub-grains Same melt pool Similar cells (case 1) Similar cells (case 2) or different cells (case 3) Different melt Similar cells (case 4) or different Different cells (case 6) pools cells (case 5) ln summary, the examples above assist in in understanding the sub-grain cellular dislocationnetwork and its influence on the mechanical properties, and demonstrate how theseproperties can be manipulated using this understanding. The features of cellular structureunder different processing conditions were predicted and proved by comprehensivemicrostructure and mechanical characterizations. At high density levels, minimizing the cellspacing increased the average hardness. The arrangement of dislocation network determinedthe fracture mode and impacted the tensile properties. A careful arrangement of continuouslonger cells was demonstrated to generate a combination of superior strength and good ductility along the building direction.

Example 2: Larger-scale samples For the as-built PV/PH specimens in Example 1, the produced specimen cross section is only1*4mm and the tensile test sample thickness is less than 1mm. ln such small specimens thedefects will play an unduly large role during the tensile tests and will result in unduly lowtensile properties. ln order to ameliorate the role of specimen size in the measuredmechanical properties, a new series of specimens were prepared (S-series). The as-built S- series specimens were 8mm in diameter and the diameter of tensile test specimens (gauge 18length part) is 3mm (polished). S-series specimens are therefore larger and defects will not beso critical to the measured mechanical properties. ln essence, S-series specimens provide amore realistic indication ofthe true mechanical properties obtainable using our additive manufactu red 316L steel.

All S-series specimens were prepared using the same steel powder as in Example 1. The laserpower used was 195 W, a layer thickness of 0.02 mm was used, and a rotation angle offset of45° was used between adjacent layers. Scanning speed and line spacing were varied as shown in Table 3 below.

Table 3. Scanning parameters and obtained mechanical properties of S-series specimens Sample Mechanical properties Scanning parametersYield strength Tensile Elongation Scanning speed/ Line spacing/Mpa strength /Mpa /% mm/s /mm S1 491,4 594,7 60,9 1700 0,05S2 521,1 599,6 67,9 850 0,1S3 521,0 614,3 82,6 566 0,15S4 536,5 601,5 79,3 566 0,08S5 535,0 589,9 83,5 700 0,08S7 516,4 583,8 68,8 1000 0,08S8 528,9 596,2 80,7 566 0,1S9 521,1 603,3 74,1 566 0,12S10 529,6 634,1 82,6 800 0,15S11 543,3 601,7 85,5 700 0,1 lt can be seen that the greater specimen sizes used in the S-series provides more consistentmechanical properties, and that the scanning parameters can be tuned to optimize themechanical properties obtained. Elongation at fracture in excess of 70%, and in many cases in excess of 80% were obtainable by appropriate choice of scanning parameters.

Example 3: Fresh powder 19 The specimens of Examples 1 and 2 were produced using SS 316L powder recycled fromprevious additive manufacturing studies, and therefore potentially comprising a substantialdegree of oxidation on the powder surface. ln order to explore the properties obtainableunder optimal conditions, a series of specimens were prepared using fresh SS 316L powder.The powder was otherwise the same as that used in Examples 1 and 2 and comprises si|icon asthe active metal. Specimens were prepared using 316L powder only (Normal), or with varyingquantities of a nanoparticle (approx. diameter 800 nm) V20; secondary phase (ODS-1 andODS-2). The laser power used was 195 W, a layer thickness of 0.02 mm was used, and arotation angle offset of 67° was used between adjacent layers. Scanning speed and line spacing were varied as shown in Table 4 below.

Table 4. Scanning parameters and obtained mechanical properties of ”fresh”-series specimens Sample Scanning parameters Secondary Mechanical propertiesphase %(W/W)Scanning Line Yield Tensile Elongationspeed/ spacing / strength / strength / / %mm/s mm MPa MPaNormal 900 0.15 0 552 661 83.2ODS-1 800 0.08 1 574 627 90.5ODS-Z 700 0.06 2 553 597 95.7 lt can be seen that under optimized conditions a steel workpiece having an elongation inexcess of 80%, a yield strength in excess of 550 MPa and a tensile strength in excess of 590MPa may be obtained. lt is though that the superior mechanical properties obtained are dueto a powder having less initial oxidation, thus providing more evenly dispersed and smallernanoinclusions, as well as fewer defects. Addition of a nanoparticle oxide secondary phase tothe steel powder may improve the mechanical properties of the obtained workpiece,especially the elongation-to-failure. Elongation-to-failure as high as 95.7% was obtained using SS 316L powder in combination with 2 % w/w nanoparticle V20; powder.

Claims (1)

1. CLAll\/IS A process for producing a steel workpiece having an elongation at fracture ofgreater than 80%, as determined by the method of ASTM Test E8-16a, the process comprising the steps: - providing a steel powder comprising an active metal; and - subjecting the steel powder to an additive powder bed fusion manufacturing process, thereby producing the steel workpiece. The process according to claim 1, wherein the additive powder bed fusion manufacturingprocess is selective laser melting, electron beam melting, selective laser sintering, selective heat sintering, or direct metal laser sintering. The process according to any one of claims 1-2, wherein the steel powder is 316L steel powder. The process according to any one of the preceding claims, wherein the active metal isconstituted of a metal alloyed in the steel powder and is selected from Si, Cr, Mo, Ti, Ta and Nb. The process according to any one of the preceding claims, wherein the active metal isadded in a quantity sufficient to react substantially completely with any residual oxygen present during the additive powder bed fusion manufacturing process. The process according to any one of the preceding claims, wherein the steel powdercomprises a nanoparticle metal oxide, and wherein the nanoparticle metal oxide is preferably yttrium oxide nanoparticles. The process according to any one of the preceding claims, wherein the steel workpiece has a yield strength of at least 550 I\/|Pa, and/or the steel workpiece has a tensile strength of at least 600 I\/|Pa, 10. 21as determined by the method of ASTM Test E8-16a. The process according to any one of the preceding claims, wherein the additive powderbed fusion manufacturing process comprises a step of rotating a scanning direction between adjacent layers. Use of a steel powder comprising an active metal in an additive powder bed fusionmanufacturing process for producing a steel workpiece having an elongation at fracture ofgreater than 80%, as determined by the method of ASTM Test E8-16a. A steel workpiece produced by a process according to any one of claims 1-8.