Sour of Material Procesing Tesh, 286 (2020) 116738 Contents lists available at ScienceDivect Journal of Materials Processing Tech. ELSEVIER journal homepage: [Link] Effect of laser remelting processing on microstructure and mechanical properties of 17-4 PH stainless steel during laser direct metal deposition ® Zhiyuan Yu, Yi Zheng, Junmei Chen, Chuanfeng Wu, Jijin Xu, Hao Lu", Chun Yu* Shona Key Labora of Mata Lave Pocesing an Mofo. eho of Maur See rd Exgering. Shanghai Tong Unive Shang 200240, ‘ela ator Ade ase direct metal deposition (DMD) i an efcent and flexible adeive manufaccaring technique which hat Keynrde ‘broad application prospects, utitis somite due to defets and mechanical anisotropy Laser emetng (LR) Inver ct meal epson Isa process that aller adh layer is deposited, eesean the deposition Iyer withthe same slice data but without ie powder deliver, nd ic is often used during elective late melting Herel, LR proces ha been applied during Reeling ‘the DMD proces of 117-4 PH tel to enhance the densification level an relieve the mechanical anzeeopy I Mechasiclaniotony {ound that the thermal history porosity and microstructural evolution are dependent on the LR energy density Ean ‘Moreover the roughness of op surface ofthe deposited layer ane intalayer porosity decease with increase of the lnuerremeling energy density. While for interlayer defects, there ls an optional LR energy density care spending to the lowest lnerlayer porosity, Puthermere, UR proves can enhance the bolling time a high temperature, even sometimes heat the sample above Ay, rerlking in change in content of aente and ca bide In addon, LR proces grety dilutes the pre-soifice tate It was also proved thatthe lt defects a plane tothe Intrayer and the anisotropy ofthe crytallographe erentatons are extremely harmful forthe ‘mechanical isotopy ofthe DMDed samples. Finally, sample manufactured wit ager remelting at 15.0/tam* strength, but the contribution of laser remelting to eliminate anisotropy in elongation is limited because ofthe Inteayer defets 1. Introduction Laser-based additive manufacturing (LBAM) isthe most promising and feasible method to fabrieate metalic parts. Laser direct metal de position (DMD) isa category of LBAM process which combines powder and energy delivery or simultaneous deposition. The DMD processing is capable of fabricating customized and near net shape three-dimen- sional component for industries like automobile aerospace and defense Fields. Taking advantages of high etfciency and flexibility, DMD ex tends application range of LBAM techniques and exploits possibilities to fabricate much larger and more complex components. Shamsaci ct al (2015) reviewed various metas and procedures in the DMD process “They summarized that laser power, laser scanning speed, powder feed rate, hatch spacing and scanning path were identified as the most im- portant faetors for final performance of materials. Huang ets) (2019) correlated the seanning speed and laser power to the microstructural ‘evolution by analyzing the localized real-time thermal characteristics. they acquired a fner microstructure by employing a higher scanning ~Corespondingasthors Email addresses: shweldatsu [Link] GH. Lu), yuehunl980stw edu en (. Yo. ops//[Link]/10.1016/)matprocec2020.116738 speed accompanied with a lower laser power. Wang etal. (2018) stu died the effect of seanning strategies on microstructures and mechan. cal properties in direct laser deposited TISAI4V. They demonstrated that island seanning srategy can improve the tensile properties by ee fining the grains, but weakened the acuity because of the con centration of pores at the island overlap zone compared to orthogonal strategy. Due tothe high cooling rate and different localized thermal history inthe DMD process large thermal residual stress, varieties of defects and anisoteopie behaviors become the critical challenges to restlet the development of this technology. Several studies have paid attention to ‘address these issues. Kim et s| (2020) applied ultrasonic nanocrystal surface modification technology to the post-treatment of DMD, and the tensile residual stress in the surface of DMDed part was converted to compressive residual stress, thereby the wear resistance was improved Contrasted to the selective laser melting (SLM) process, DMD process can induce more pores, cavities and lack of fusion with the increase of Tayer thicknesses. Wang et al, (200%, ) investigated the pore Received 26 February 2020, Received in cevsed form 1 May 2020; Accepted 2 May 2020 ‘Available online 11 May 2020 992440136/ © 2020 Fleevier BV. Allright reserved.Jol of Marrs rca Tech 286 2020 116738 rable? ratte {Bente coupes of 17 por Bete prceing praetor den of meinen © « oN a « wf fom le oe Seming tory Reig Free Miting Senay Sty Se ty a a a won pny Srey Sra formation in DMD proces sng diferent metal powders. They divided EMDURI «8S ab " imma detects into two categories itroyer nd inalayer Reyes MDUNZ SEE ie found that interlayer pores correlated with oxidation kinetics andi tralayer pores were sensitive to scanning speed, Wollf et al. (2017) utilized an in-house thermal simlator to link localized cooling rate with porosity geometry in adlitive manufactured titanium alloys. They ‘ound thatthe interlayer defects prefered to appear at low input energy ‘condition, meanwhile, the intrslayer defects preferred to appear at re lative high input energy condition. Bandyopadhyay et sl. 2019) i vestigated the effect of deposition orientation on fatigue performance of DMDed TI6AI4V. I indicated that the fatigue life could be improved by 107/90" hatching process, some of the pares were filled in by successive passes at diferent orientations. Additionally, regularity of void dis tribution and existence of textare induced a dramatic ansoteopic be havior. Qiu etal. (2016) investigated anisotropic behavior of Invar 36 ‘components in different built direction. It was found that the inter ‘granular falure mode enhanced strength in the horizontally built samples while a transgranular failure mode presented inthe vertically built samples with poor ductility, Beside investigations on the process Parameter optimization of DMD, several rescarches focused on further heat reatment as mentioned by Cheruvathus etal. (2015), hot isostatic pressing as mentioned by Riemer eal (2014) and remelting to reduce Porosity eliminate anisotropy, and uniform microstructure. Laser remelting (LR) ie a convenient and efficient accessory process {for the LBAM, that seanning again on every solidified layer without powder deliver. This process supplies chance for pores to escape and makes porosities decrease, Meanwhile, LR can uniform the composition and eliminate residual stress. Wei ct al. (2018) studied the influence of ‘the repeated usage of LR process on deposition quality duing the SLM processes of the [Link]. I was demonstrated that LR treatment Induced two kinds of preferential orientations and multiple cycles of LR aes ae process reduced the residual stress. Yasa (2011) applied LR process on AISI 316L parts using different parameters. It was found that LR with loveenergy input ean dramatically improve density of the samples, nonetheless, increasing LR cycles had a lesser extent on density im provement. Moreover, Chen et al. (2018) developed "SLM plus LR" {echnique to obtain # complete lath martensite microstuctute in tool ‘tel. was found that LR process led toa higher cooling rate in powder bed system and generated ultrafine lath martensite, resulting in en- hancement of hatdnes, Yang et. (2018) investigated the inflence of LR process on the magnetic properties of Fe-Co-based alloys during ‘SIM, Results showed that the secondary phase was produced in the recrystallization process after LR and the alloys exhibited higher so ‘uration magnetization. ‘The 17-4 precipitation hardening (PH) stel is an attractive metal for LBAM due to its good weldabilty and austenitie/martenstie mi crostructure as mentioned by Nezhedfar etal. (2019), which has been used for components and structures in various fields like nuclear, aerospace, marine, naval and chemical industries as reported by Kumpaty etal (2013). Alnaljar etal. 2019) proved that asfabricated 17-4 PH samples hada fll ferritc microstructure and exhibited a grain epitaxial solidification crossing several layers due to the high cooling and heating rates during SLM (10°-10°K/s), The large thermal gradient led to a strongly anisotropic microstructure as well, Carneiro et al. (2019) analyzed the fatigue and cyelic deformation behaviors of ed dlitve manufactured 17-4 PH, I¢ was observed that defects like pores were the main cause for low fatigue endurance limit. Furthermore, Schaller etal. (2017) explored the corrosion susceptibility of LBAMed fast roster ee Z Fig 1. (a) Schematic fr DMD feilty (6) DMD scanning strategies (QDMD + IR scanning strategies.zreat Jol of Marrs rca Tech 286 2020 116738 fam gee Fig. 2 (4 Schematic showing tensile specimen positon (b) Dimensions of tense specimens Cuding direction, scanning direction, hatching direction) “GGSGS DMD#LRI DMD#LR3 Fig. 8. Top surface LSCM images andthe measured vowel sample, ‘hice: 4mm on (-e Laser remeting 2 8:7 rr? i intense fg > Laser copostion at 657 Jr? eo 3 wo 2 ao g fd E200 00020406 08 149 Zen) Fig 4 (2) Thermal images and () Temperature gradient long the Z cetelines ofthe molten pool under diferent processes, 17-4 PH steel. And electrochemical measurements demonstrated that reduction of the corrosion resistance was attributed to the porosities ‘with diameter =50ym. Whereas, lite research has considered LR to enhance DMD process, as a result, the effects of this novel approach on microstructure, porosity and mechanical anisotropy are not well un derstood. In this paper, 17-4 PH steel has been prepaced by DMD, combined with different LR process, and the microstructure, thermal history, porosity and mechanical anisotropy have been investigated empensnre °C) 2. Materials and experimental procedure 2.1, Materials and DMD process Gas mOE Tibe [aint une sonra asso ssa The experiments were performed with a gae-atomized 17-4PH steel Time powder supplied by JTUCHUN (Suzhou)Co, Ltd. The particle has a near spherical shape and a diameter range of 70~ 150m. The chemical compositions ofthis powder are shown in Toble 1. The 304 austenitic stainless steel was chosen as the substrate with a dimension of 4100:mm % 100 mm % 20 mm. The experimental facilis are shown in Fig la. A si-axis robot (kul, keB0ha) equipped with a coaxial nozzle (@rectec, ¥CS2) was employed to realize the designed DMD paths. The Fig. 5. Thermal history of one layer afer soliifiation under dierent proareal Journal of Marra: rcs Tach 284 020) 116788 Teal porosity erayer porosity Hi Laatsyer porosity DMD DMD+LRI DMD#LR2 DMD#IR3 ‘Sa Fig. 7, Porories and relative densities of samples subjected to diferent po laser radiation was provided by IPG laser (YLS-2000-CT), which de livered a maximum output power of 2kW with a wavelength of 1070nm (laser beam diameter = [Link]). An InfraRed RSOOEX-Pr0 ‘camera was used to monitor the thermal behavior during the prepare tion process. The camera captured three frames every 1, then the ‘thermal profile ad the temperature ofthe melt pool were acquired by InfReC Analyzer software. The emissivity of 17-4 PH steel were cali- brated for temperature measurement with thermo-covples, The pow. ‘ders were supplied with an argon gas feeder and shielded by argon gas. ‘A series ofcuboids with dimensions of 20mm x 50mm x 40mm have been deposited ‘Based on a series of preliminary experiments, the optimized pars meters (laser power of 685; seanning speed of 10mm/s; powder feeding rate of 7 g/min; shielding gas rate of 15 L/min; layer thickness ‘of 0.5 mm; hatch spacing of I.1 mm) were obtained and used for direct ‘metal deposition. The bidirectional strategy with a 50% dislocation between central line of adjacent layers was employed, as illustrated in Fig. 1b. And schematic for “DMD plus LR” process is illustrated in Fig le. For convenience, the preparation conditions without LR treat ‘ment would be refered to as “DMD”. When it comes to the conditions ‘with LR treatment, the diferent energy densities for LR processes are listed in Table 2. Meanwhile, the corresponding samples would be re ferred to as "DMD + LRI", "DMD + LR2", and “DMD + LR3”. In a Aiton, the energy density, E was calculated by using Eq. (1) a8 sug ested by Toyserkant et al. (2008), @ ® ‘where, Pi laser power, vis scanning speed, and d is laser beam dia 2.2, Surface observation, porosity analyses and microstructure characterization ‘A Laser Scanning Confocal Microscopy (LSCM) was applied for the etal of the top surface topographies, and the roughness (Ra) was ‘measured. The volume faction of global porosities (relative density) was ‘measured using the Archimedes method a proposed by Spierings (O11), The measurement was repeated three times, with & measured “density of 7.80 g/cm’. Samples were cut from the center of transverse section (7 plane) of the as-deposited workpieces, then they were ground and polished. A solution of 4 cupric sulfate, 20m, hydro: chlorie acid and 20mm ethanol was used to etch microstructure. The distribution, shape, size of internal defects and microscopy were ans lyzed by Carlzeiss AxioCam-MReS optical microscopy (OM). For each sample, fifty pictures taken at differen locations of transverse section (OG-2 plane) were considered to estimate the porosity level, TESCAN ‘MIRA Scanning Hlectron Microscopy (SEM) was used for micro structural observation, accompanied with an Oxford Aztex X-MaxN8O Bnergy Dispersive Spectrometer (EDS). In addition, electron backzr eeat Jol of Marra rca Tech 286 2020) 116798 Fig, 8. OM and SEM images showing the microstructure of the as-fabriated samples (a) DMD ;(, €) DMD + LRL: (e, | DMD + LR2; 8) DMD + 183,rable 3 EDS analyses results coresponding to Fig 8b (a0. ‘Table 4 Comparison ofthe mechanical properties of samples under diferent processing ‘Sumples eat) Uae) BOD scattered diffraction (EBSD) mapping was performed to obtain crystal structare information (accelerating voltage: 20k, step size: 0.3m) of a certain layer in the middle of the samples, the samples were elec twolytcally polished at 12V with electrolyte solution (10% HCIO, + 909%6cH,04). 2.3. Mechanical property anelyses ‘The tensile specimens were prepared by using wire electrical dis: ‘charge machining (EDM), along the horizontal and vertical directions, asdefined in ig. 2(a). The dimensions of tensile specimens ace shown Fig.26). Tensile teste were conducted at room temperature by em ploying a Zwick/Roell 2020 machine (stretching rate = 1 mm/min). For every direction, three tensile samples were tested. To further ob- serve the fracture surfaces, SEM wae used. 3, Results and analyses 3.1. Swface observation ‘The LSCM images are exhibited in Fi. 3. I shows the top surface morphology and the measured roughness (Ra). In the DMD sample, content: 0.82% OTN Fig. 9. EBSD phase fraction mapping for samples fabricated with (2) DMD; (b) DMD 1 Jol of Marra cen Tech 286 2020) 116798 many adhesive particles and obvious scanning traces can be observed fon the top surface, resulting ina high Ra, 107 um. After LR processing, the number of adhesive particles is significantly reduced, and the cor responding Ra reduces to 97m for the DMD + IRI sample. In the DMD + LR2 sample, the surface becomes even much smoother and neater, the Ra value is measured to be S6um. Further, the scanning traces on surface are almost incistinguishable in the DMD + LR3 sample, aad the Ra i only 17 um. 3.2, Thermal history The in situ thermal images of the melten pool during laser depos ing and LR processing are shown in fig. (a). The boundary of the rolten pool is distinguished by considering the isotherm of tne melting temperature (1440°), and the Z direction is defined as the building direction, the ¥ direction is defined as the laser scanning direction. The depth of laser deposition molten pool is approximately 653 ym, After the LR treatment, the depth of molten pool decreases to 231 um, 147 un and 104um, corresponding to che energy density of 8.7, 15.0 and 21.6.4/mmé, respectively, ig. (b) shows the temperature gradients along the Z centerlines from the molten pool surface to the solidified part of each pool. The laser energy density significantly affects the emperature gradient and It canbe seen that the maximum temperature gradient is closer to the surface with the decrease of energy density ‘When the laser remelting energy density is 87 J/am?, the maximum temperature gradient reaches a maximum value, about 2500°C/mm, ‘ehich results from the minimum depth of pool and rapid heat dis sipation atthe surface The workpiece is continuously heated during the processing. Fig. 5 shouts the thermal history of «layer inthe center of the workpiece after solidifeation, where the influence from substrate can be neglected. is observed that the LR process prolongs the heating time. Austenitic wansformation-starting temperature (Aej) of the 17-4 PH is 827°C a8 investigated by Bhacir et sl. (1999). The deposition of next several layers heats the sample above Ac, and the holding time above Ac lasts Tongest in the DMD ~ LR2 sample, With the layer by layer deposition, the heat dissipation is basicaly the same as the heat input, and tem: perature tends to stabilize. At lst, the temperacure drops rapidly as the ‘manufactaring process finished, ace aes (o Dap + 1R2; 4) DMD +R, indicating the y-Fe(ed) inthe Fe rmawxtbue) Gor interprecaion ofthe references to color inthis figure legend, the reader s referred to the web version ofthis article)Jol of Marrs cen Tech 286 2020) 116738 boundary 3.3. Microstructure and crystallographic orientation characteristics 3.3.1. Defects characterises ‘The representative internal defecs ofthe four samples are shown in ‘ig. 6, The (ypical defect in the DMD sample is the unmelted particles between the two layers as shown in Fig. 6(a). The lack of fusion be ‘comes the dominant defect in the DMD + LRI sample due to the low. lying area formed by scanning trace, as shown in Fig. 6(b).Interlayer defects are significantly decreased in the DMD 4 LR2. sample, as shown, in Fig. ), but a few pores still exist inside the ayers. While, the i terlayer defects increase obviously in the DMD +13 sample. Ac cording to Fi. 5d), triangular un-melted areas appear on the straight {sion line, which symmetrically distribute on both sides of the poo. “The porosity and the relative density ofthe samples are counted, and ‘the results are shown in 'g.7-In the DMD sample, the measured total porosity is around 0.82% (interlayer porosity: 0.45% intalayer por ‘sity: 0.37%). While the total porosity in the DMD +LRI and DMD + LR2 samples become lower, 0.56% (interlayer porosity 0.20%; intalayer porosity: 0.3986) and 0.26% (interlayer porosity: 0.13%: i ‘walayer porosity: 0.138), respectively. However, the total porosity in the DMD + LR3 sample obviously increases to 1.37% containing more interlayer defecs (1.298). As far asthe relative density, as plotted in Fig?, the DMD + LR2. sample exhibits a highest relative deasity of 99.06% = 0.5886, while the DMD + LR2 sample has the lowest relative density of 98.07% = 0.61%. 3.3.2. Microstructural observation ‘Tig. 8 shows the microstructures observed by OM and SEM. The OM. mages show that a series of half-ellipses structures display in the ‘transverse sections of samples. The mean height of layer in the DMD sample i $60 um, as shown in Fig. S(a), that in the DMD + LRI sample is 557 um, as shown in Fig. 8(€). But the mean height of layer in the DMD + 1R2 sample reduces to 504m, as shown in Fig. 8). Pari- ccularly, the DMD +LR3 sample has a typical double halfellipses structure with 603 um in mean layer height, as show in Fi. 8(g). The microstructures of the four samples are shown in Fig. 8(), (@, (0, (D), ‘which consist of dominated martensite and &ferrite with’ varios ‘morphologies. The 17-4 PH steel as a feriti solidification mode (FA) 88 proposed by Bahram Balajaddeh and Naffakh-Moosavy. (2019). This alloy solidifies as a primary ferrite during the inital stage, then, the ferrite mostly transforms co austenite during the subsequent cooling process, However, the cooling rate in the DMD process is particularly 10, SD Inverse pole figures Z fr samples fabricated with (a) DMD; (b) DMD +121; (© DMD + LR2;(€) DMD 4 LRS and the black lines reveal grain pid and the transformation is partially suppressed. As the temperature Arops to room temperature, austenite gradually transforms fo marten. site, and retained Serie is expected in the martensitic matrix. There exists vermictlar, dendrite and lathy ferrite in martensite matrix. The ferrite morphology might be related (0 the Cray/Nigg ratio and the so lidfiation/eooling rate, which was considered in previous study Gilmes e¢ sl, 1996) ‘Additionally, a great quantity of precipitates is found according to Fig.80), (), (, (). The precipitates are identified as carbides ac cording to the EDS results sted in Table 3, The carbides formed in 17-4 PH steel mainly consist of MysCe as reported by Sun et al. (2018) and Philips, 2015. Since the volume fraction of precipitated carbides in the rmatex is dificult co be obtained directly, the topographical effects are ‘ignored, then the volume fraction of carbide particles was messured from Fig. 8(6), (@, (O, GH) using Image) software as developed by Schneider etal (2012). The volume fractions ofthe carbides are 0.375%, 0.40%, 0.63% and 0.24% in the DMD, DMD + LR1, DMD + LR and DMD + LR3 samples, respectively. Generally, cazbide precipitations in the additive manufactured 17-4 PH increase with the increase of tem: perature and extension of holding time during heat treatment, a in vestigated previously by Cherwvathur etal, (2016). The LR process changes the volume fraction of carbides by changing the thermal his tory of workpiece, this will be discussed inthe later seetion Table 133.3. Orytalographic orientation analyses EBSD phase maps for the four samples are presented in Fig 9. AB can be seen from Fig, Sa, the volume fraction of austenite i only 0.82% in the DMD sample. Comparably, after performing LR process with a low energy density, like the DMD + LRI sample, the volume fraction of austenite is basically unchanged, itis 0.86%. While asthe remelting energy density increases, like the DMD + LR2. sample, the volume fraction of austenite increases to 7.79%, However, the austenite frac tion reduces to 3.65% in the DMD + LR3 sample, ‘The EBSD inverse pole figure maps and the rain boundary maps of the four samples are shown in Fig. 10. These maps show a conventional martensitic block structure. The grin size in the DMD sample, around 1.25 um, is obviously coarser than the others (0.88 ym for DMD ~ LR; 0.80ym for DMD + 1R2; 0.81 um for DMD + URS). LR processing produces a smaller melt pool on the surface of solidified layer compared ‘with the melt pool produced directly by laser deposition, then leads toa higher cooling rate and refines the crystalline structure as investigated by Chen etal, (2018), The corresponding pole igure maps are shown inzr eeat Jol of Marrs cen Tech 286 2020 116798 100} alt} @) ax = 8,67 @: () % © max =3.37 iL @ & B i Fig. 11 HBSD pole figures fr samples fabricate with (a) DMD: (6) DMD + 181; (€) DMD + LR; () DMD + 1. aro} ®r200 ‘0 100 : So gem i | feo oa i'l x) —ow Bebe a eames) am busts 02 4 6 8 Ow 6 o 2 4 6 8 0 2 4 1 1 cea oni Fig, 12, Representative tensile stress-strain curves of samples in diferent processing () Horizontal (6) Vera. Pig, 11, The DMD sample has a random exystallographie orientation, the DMD + LR2 sample has the most chaotic erysallographie orienta ‘with a maximum polar density of 8.67, as shown in Fig. 11(a). The LR ons, with a maximum polar density of 3.37. ‘weatment evidently affects and further disorganizes the original crys In summery, LR process with appropriate energy density ean de tallograpic orientations a shown in Fig, 1Ib-d. The maximum polar crease the intralayer and interlayer porosity, so that the sample has a ‘density in the DMD + LI sample conspicuously reduces to 421, and high density. Meanwhile, LR process changes the thermal history, andJol of Marrs cen Tech 286 2020) 116798 Fig. 18. SEM micrographs showing the fracture surfaces of horizontal samples (2, b} DMD; (,d) DMD + LRI; (e,.DMD + LR2; (5) DMD + LRS. the local temperature gradient of the molten pool, further influencing the pre-solidified texture the volume fraction of precipitates and aus- tenite, These changes would play an important role on the mechanical behaviors. 3.4, Mechanical properties ‘The representative room temperature tensile stress-strain curves ‘along both horizontal and vertical directions of the as-fabricated sam. ples are presented in Fig. 12 and the ultimate tensile strength (UTS), (0.2% proof stress (a3) and break elongation (ED are summarized in Table 3, On the whole, the horizatal tensile samples are found to own not only the higher tensile strength but also much better elongations compared with the vertical tensile samples. [As for the horizontal tensile samples, the UTS improves from 1129 + SMPs for the DMD samples to 1223 = 10MPa for the DMD + LR samples, while the elongation is basically unchanged (rom 140 + 0494-141 = 0.396), As the LR energy density increased, the UTS drops slightly to 1085 = 9MPa (for the DMD + LR2 sample) and 1105 + 8MPa (for the DMD +LR3 samples). Corresponeingly, the yield strength reduces to 542 + 16MPa and 524 + 32MPa, and the elongation increases to 16.6 + 0.9% and 15.2 = 0.286 in reverseJournal of Marra: rca Tech 284 20) 116788 Fig. 14. SEN merograps showing the fracture surfaces of vertal samples (,b) DMD : ( &) DMD + LR; (, DMD + R28) DMD + LS. Fig 12 presents the fracture surfaces of the horizontal samples. Many squeezed and elongated defects present in the fracture surfaces, they are associated with ellipsoidal or triangular unfused voids, as shown in Fig. 15(@), (©, (@, ()- While the DMD + LR2 samples exhibit rela ‘ively less and shallow defects, corresponiling to the minimum porosity According to Fig. 130), (), (D, (h), a fairly ductile fracture mode is ‘characterized by the presence of large quantities of fine dimples. [As for the vertical tensile samples, the UTS reduces by 18.7%, (0 ‘918 = 13MPa, compared with the horizontal tensile sample in the DMD sample, and the elongation suddenly reduces to 3.8 = 1.3%. As for the DMD +LRI and DMD + 1R2 samples, the UTS increases to 982 © 11 MPa and 1073 «= 12MPa, meanwhile, the elongation in creases 10 43 = 0.4% and 5.6 + 0.6%. Whereas, the DMD + IR3 sample fractured at 575. 7MPa with basically no elongation, becase ofthe quantities of interlayer defects (Fig.7). It should be noted that the UTS of the DMD + LR2 sample along the horizontal and vertical di rections is almost identical ‘The fracture surfaces of vertical samples are shown in Fig. 14 ig. 24(@) shows some open-up pores and pockets of un-melted powder particles on the fracture surfaces ofthe DMD sample. And = mixture of Auetile and brite fracture mode is exhibited, as shown in Fig. 1400). ‘This pocket of unmelted particles is signifcandy reduced in theJol of Marra cen Tech 286 2020) 116738 Tagen Laserremelting _Layerntl none MDALRY(== Yas 7) eee GE © pow Ce MORI) ca x ——mettpool boundary Fig, 15. Schematic of delets formation during diferent processing conditions oot 14] 165 168 1e40| 1e2| Volume fraction of M,C, (6) 144] ‘0 1000” 2000” 3000” 4000 Time (6) Fig. 16. Relation Between the fraction of MayGy carbide precipitation and heating time at eitferent temperatures. DMD + IRI sample, as shown in Fig. 14@9, but chere are still brittle areas on the fracture morphology, as outlined in Fig. 14(@). According to Fig. 14(@) and (D, a great quantity of deep dimples present in the DMD + LR2 sample, meantime, the interlayer porostes still exist, so the ductly is limited. Especially, in the DMD + LR3 sample, large ux- melted areas (as shown in Fig 5()) act as crack initiation sites for brittle fracture as they ae vertical to the tensile loading direction. This phenomenon was also explained in the work done by Lebrun et a. 2014, 4, Discussion ‘The impact of remelting energy density on layer morphology and defect distribution are illustrated in Fig.15. After laser deposited, un melted powders stick to the surface, and clear seanning tracks appear fon the surface induced by the melt pool wettability as considered by ‘Leung etal (2018). Meanwhile, some pozes remain inside the samples, because the argon shielding gas and the gas contained in powders are entrapped during solidification as suggested by Sinestha etal. (2019) ‘When IR process is performed with the energy density of 8.7.1/mm*, the unmelted powders stuck to the surface are eliminated, surface roughness and interlayer porosity reduce significantly, but the in walayer porosity changes very litle, As the remelting energy density increases to 15.0/mm’, the intrlayer porosity decreased due to the Aeeper penetration and higher possibilty fr the trapped gas to escape, which was also considered in another work (Leung eal, 2018). Parther Increasing the remelting energy density to 21.6J/mm4, the inralayer porosity decreases to 0,08%, but the interlayer porosity increases dra matically. The heat loss of the workpiece is primarily through surface [Load Direction! sipplane: (110) (Ti Loa Direction /'Y_ slip direction: <111> DMD*LRI “wll POL TE ee fl DMD+LR2 | 9F Dab 1 | Schmid Factor Fig 17. HBSDaderived (110) < 111 > Schmid factor valuesvokime faeton ofthe sample.convection and radiation during the manufacturing as considered by Wang eta. (200%, b), and the smooth surface has stronger convection as it undergoes the blowing of argon shielding gas. This inference is ‘consistent with the phenomenon thatthe temperature ofa layer in the DMD + LS sample deops as low as 250°C after solidification, which s lower than other processes, as shown in Fig. 5. Moreover, A rough surface causes multiple reflections ofthe laser, and tiny grooves fail itates laser radiation transfer, improves laser absorptivity, thereby the reduction of roughness will decrease laser absorptivity as mentioned by ‘Wang etal. (2000). When the next layer is deposited, on the one hand, the lower surface temperature will increase the viscority ofthe droplet making the liquid metal dificult to spread out as implied by Kumar anc Prabhe. (2007), On the other hand, eduction of absorptivity weakens the penetration of melt pool and the overlap between layers becomes thin, ‘The cooling rate in the molten poo, as well as the thermal history at ‘each position of the workpieces are believed to determine the micro structure development. The critical cooling rate for generating over 99% martensitic in 17-4 PH is 10K/s as mentioned by 1sia0 et al (2002). However, Huang etal. (2019) reported thatthe cooing rate of DMD process can reach above 1000 K/s, and this high cooling rate re striets a fewr transformation of Bferrite co [Link] in reverse ac ‘ording t0 Das etal. (2006). Besides, the additonal laser remelting, process changes the thermal history of the sample. For DMD + LR sample, the average temperature ofthe certain layers from the center of ‘workpiece is 475 °C and lst fr 5000s. While for the DMD + LR2 and DMD + 1R3 samples, the mean temperature is 630° and 350°C, re spectively. The relation between the volume fraction of precipitated MasGe carbides and the holding time at different temperatures is cal culated by Jmatpro software as developed by Saunders et al. (2002), as plotted in Pig. 16. It ean be summarized that the volume fraction of ‘MayGy increases remarkably as the temperature Keeping at 630°C for 50005. This phenomenon is in agreement with the increase of carbide ‘contents in the DMD + LR2 sample. Moreover, the austenite volume fraction increases in the DMD + LR2 sample because ofthe longest time shove Ac;, leading more unstable martensite revert to austenite. The solulfcatlon microstructure is determined by temperature sradient and solidification rate az mentioned by Kurz etal. (1986). In sddition, the temperature gradient and solidification rate vary sig- nificantly throughout the melt pool depth by several orders of magni tude as reported by Hontha etal. (2000). As it ean be seen in Fis. 4, different LR processes generate diferent melt pool depth and tem perature gradient. Redistribution of the temperature gradient can realy dilute the pre-solidified texture, thus lesding to a very smaller polar density. I's worth noting that subsequent melting of an overlying layer can dilute the texture again. This explains the phenomencn that polar density ofthe DMD + LR3 sample with the weak overlap between layers has a slight increase compared to DMD + LR2 sample. In all preparation conditions, beter properties are obtained in the horizontal direction than in the vertical direction of the samples, re sulting in a pronounced anisotropy in mechanicel properties ofthe laser direct deposited 17-4 PH. The FRSD-derived Schmid factor v2 Iues—volume fraction of the samples are plotted in Pig. 17. For DMD sample, only 49.4% grains have Schmid factor value ranging from 0.4405 in the vertical sample (load direction // 2), wile ti 67.5% in the horizontal sample (oad direction // ¥). Tis diference indicates that more grains favorably oriented for slip acivstions in the horizontal sample, contributing to local plastic deformation, AS LR process is performed, the difference in this pereentage between the horizontal ané vercieal samples is within 1.6%. Ie ean be inferred that LR process minimize the impact of crystallographic texture on mechanical aniso ‘ropy. Especially for elongation anisotropy, it's strongly affected by the presence of porosity as mentioned by Lebrun etal. (2014). It ean be analyzed thatthe elongation for vertical samples is reduced with the Increase of interlayer porosity, but there is no correlation between Jol of Marrs cen Tech 286 2020116738 elongation and interlayer porosity for the horizontal samples, such as the DMD + LRS sample has both high interlayer porosity and elonga- Lion, The reason is that the fat defects inplane (o the interlayer are ‘vertical to the loading direction, and easier to tear, further resulting in ain open-tear fracture made (1a), (9), (€, (g)). Whereas, a8 the in plane defect ae parallel tothe loading direction, deformation leads to Shear-stretch mode, exhibiting squeezed and elongated defects (ig13@), (©, 9, (). Regardless ofthe distribution of defects, aus tenite transforms to martensite during plastic deformation, and this stress-induced transformation reduces yield strength, along with in crease of work hardening rates, and lengthens stage of wniform plastic deformation according to LeBran etal (2015). 5. Conclusion In this study, diferent laser remelting parameters were employed during the DMD process of 17-4 PH steel, and their effects on porosity, thermal history, microstructure and mechanical properties ate sys. tematially analyzed. The primary conclusions are drawn as follows; (1) LR process can decrease the porosity level in the DMDed specimen, ‘hich is beneficial in decreasing the mechanical anisotropy. (2) LR process exhibit a remark dilution effect on the pre-slicified texture, minimizing the impact of erysallographle texture on me. ‘chanical anisotropy. (6) Sample manufactured with laser remelting a¢ 15.0,3/mm* has the lowest porosity and near isotropous UTS, While, the contribution of [ER process to eliminate anisotropy in elongation is limited because of the inevitable interlayer defects (CRedif authorship contribution statement Zhiyuan Yu: Investigation, Methodology, Visualization, Writing original draft. Yi Zheng: Conceptualization. Junmei Chen: Conceptualization. Chuanfeng Wu: Conceptualzation. Jijin Xu: Conceprualization. Hao Lu: Funding acquisition, Writing - review & eiliting, Supervision, Chun Yu: Punding acquisition, Writing - review & editing, Supervision, Declaration of Competing Interest ‘The authors declare that they have no known competing financial interests oF personal relationships that could have appeared to intl ence the work reported a this paper. Acknowledgments The author would acknowledge the supported by the National Nature Science Foundation of China (Grant No. 51575347) and National Key RED Program of China (Grant No, 2018VFA0702805) References Bundyopdiyay. A. ea 2019. tacece of deposition erintatn tg tespose ‘eed tal on prope welded jis Wed ne IO (10), 297-408 hie‘esrucreinbeamsbued etn of aly 3D sacar Mate Set ng AST 514 s41-3t8 pel ory 10106 msen 200902019. ‘came, Lea, 2019, Cyl éormaton an cg bar of ey ma Tacired 17-4 PH anes steel It 3. Faigue 125, 22-30 hon /[Link] 1010167 ‘che Iya 2018 paved pprnch a iet preparation of complete lath mares Theatre ota sel ysl laer meng Matra 22, 196-31 ronda ore/10 20.6 mater o1a05 0, ‘urate, 2916 hae ansfactring of 17-4 PH sls sel: pose Exstosar nips dango uvnatmaionses, Dag het ay 206 Welly of 17-4 sane ln avraged heat rete ‘ood, Teh Wed Jn. 115), 502808, ipso 110179) Yeah Indie tie denny ole somes onetorjotcoe ion parmesan metre In hse iced eet deperton (ude im ES, ot al, 2020, Uv noc ufce mdieation of highapee ol ‘tea ls M0 hpeced in ect nay deporte, Mater Poser Teena 277 Tet. hae eg/TO 1S Ianto 2091s ua G. Pra KN. 2007 Review of aonseate al eecve wel fg on ener A. Cal intertce S158 (2) l-89ip ewe 10 2018 amt, $, Kamar 5, Tomi, 8, Yo. Kamel, Andes, Govndra, M aosang,N alfbramanian Ky 2013 et feat ene’ on mechanical? Potalrge roperies of det etal laser sere 17-4 recite arene Teta eal 36 (428-800 hts ido ng sorone eovrevence ose ebro, 2016 Sean rate ses end mechs asovpy of lec ser Teed 174 Panes el Meh. Erg 71 (5) SMMDOA® ip o/To teung CLA, etal 2018 est ay img of dle and an poe raisin oer ative marntactring Nat Common 9 (1), 1355 hap canoe 10038) echt, PD. et al, 2019, Fatigue eat rom behavior of atively manacaed Tra stllesselcte ol erenation and [Link] 2. ley Son. i, Hoboken, RE. ord ar/lo Loon v7sl1iSto\sv7 eh ‘Cy weal 26 See Iver meting ova 3: erate ane proper, Jol of Marra Proce Tech 286 2020) 116798 ‘manfactred by selective laser melting. Eng. Pract, Mech. 120, 15-28. heel Siro ole engcmech 91603 086, Sounders ea, 2001 Ane computer program er predicting mater properties. Phare guia Dit 22 O59. hip! /ea-eg 1D el ‘Sehneder, CA. Rasband, W'S. lc, KW, 2012. NM image to iauged 25 yeas of Teagan Nat Methods 9 (7), 67875 enero Lose amet 089 Tactrig: Pat rebar! bebe, prses parame aptiatin and son ‘hic Mana 8, 12.85 nigoy deorzs10 oleh scans 21507 oe ‘Stern, Rat al. 2019 a avenge specie pope t pat peseance Mano 2910407, hp /dererg/t0101o}stamaTs 0807 Spiers, AB 2011. Campanton of en mearremer tehigus for ave Toeufeted metal parts, Rapid Pte. 17 () S80-986. spe te Sum ¥. La 2018 fe feat ete on arora eveuin of atv ipertotor/a0 Tole) mater 20807 38, " “eperan Ropar, A Stphe,F, 205. ater lading Pot. SPE Int So. Or ng 11), S85-092 Nl ere/10 2351/1 21858 ‘Sheri i ser sre eatent. J Appl Pips 87), 4245-358 Hs cra ioesn arth Wang, eal, 200% xperinestal andnaneraldy of he ENS apd abrcton cet Mane fg 999), O40}. hig eo oe 1011 T7995 Wore Pr PF Radi ey. 3, Nang PT. Hameo Mi Seng 131 (5), 05108. epe/dovrg ios Thebes popes of det nsedeposteg THOATAV street Metall Sn 32 0), 2175-1180, Miss /d ore 10 1007 to TosaT 483. Wei way 2018 Beto ae emeling a deporon Qual, roa et, mi “esr, an meth propery of ec er aking red Wl $2, 2017. tmework tk load coling and prope of iret erty deponten (DED) roceseé THAAD. Aa Maes 182 100-117 si ‘Yang, tal, 201, ier remeng on miro and magne prope of Fe ‘GoSard alos produce by le alae marae, [Link] Chen S613, ‘Yor 2011 The nett fhe nvuene of ser reeling on deny surge Spiga nec ea ng pt td rs 57