1 Introduction

Inconel 625 is a nickel-based solid solution-strengthened superalloy that exhibits excellent mechanical properties and corrosion resistance at high temperatures. Therefore, it is commonly used in aerospace, marine, chemical, and power generation industries. The nominal composition of Inconel 625 is: 20 to 30 pct Cr, < 5 pct Fe, 8 to 10 pct Mo, Nb + Ta in the range 3.15 to 4.15 pct, < 0.1 pct C and < 1 pct Mn, Si, Al, Ti, and Co.[1,2,3,4]

Nickel-based superalloys are characterized by poor machinability and high work hardening rate, leading to rapid tool wear and decreased in work life.[4,5] Therefore, additive manufacturing (AM) methods of superalloy parts are increasingly being developed to reduce machining and obtain complex shapes with internal channels.[6]

AM is a dynamically developed group of processes used to produce three-dimensional objects layer by layer from a Computer Aided Design (CAD) model.[7,8] Nowadays, the most popular process that allows fabrication of metallic components with good mechanical properties, low porosity, and high accuracy is laser powder bed fusion (LPBF).[9] Currently, the LPBF process allows the production of metallic parts with almost full density and mechanical properties similar to those obtained by conventional methods.[10,11] However, in the LPBF manufactured parts some imperfections in the form of pores and cracks may occur.[8,10,11,12] The absence of fusion defects can significantly extend the useful life of AM parts.[6,7] Due to the intensive development of AM technology in recent years, a lot of efforts are associated with the designing of new materials, optimization of the processes as well as post-processing to obtain optimal material properties, strike a balance between part quality and process efficiency, in situ control of process,[9] multi-materials production[7,13] and use of multi sources of energy.[13]

The formation of discontinuities and impurities in LPBF parts made of different metallic materials is difficult to predict, and even a small number of large lack of fusion (LOF) pores can significantly reduce its mechanical strength and lifetime.[14] It is well-recognized that AM parts may be more prone to fatigue failure due to the presence of large pores, which are sites of stress accumulation. In as-deposited condition, LPBF Inconel 625 might be almost fully dense. The porosity depends mainly on the volumetric energy density (VED). By parameters optimization Marchese et al.[15] demonstrated that residual porosity in LPBF Inconel 625 might range from 0.025 to 0.10 pct. Meanwhile, Pleass and Jothi obtained porosity below 0.15 pct for two different powders.[16] The porosity is also influenced by other factors, such as the laser spot size[17,18] and gas flow.[19] Sow et al.[17] reported that the 0.5 pct and below 0.02 pct porosity in LPBF Inconel 625 was obtained with laser spot diameters of about 80 and 500 µm, respectively. They postulate that higher porosity Inconel 625 LPBF processed with lower laser spot diameter is related to the lower stability of the melt pools and, thus, a higher rate of spatter formation. Also, Koutiri et al.[18] observed in LPBF Inconel 625 the porosity in the range of 0.12 pct to 6. They found that using a 160 µm laser spot diameter resulted in significantly lower porosity compared to an 70 µm diameter, while keeping all other parameters constant. Reijonen et al.[19] showed that the porosity in the LPBF processed materials depends also on the protective gas flow resulting from the machine design.

The recommended treatment to improve the density of the AM parts is hot isostatic pressing (HIP), which allows to remove most of the pores in the LPBF processed materials. The HIP is often used for LPBF Inconel 625 as-printed parts, but during this post-processing treatment the material microstructure changes due to recrystallization and grain growth, thus causes deterioration of the tensile strength,[20] but provide to reduce the porosity of Inconel 625 below 0.1 pct.[21] In addition, HIP is time and energy consuming, therefore makes the manufacturing process complicated and multistage, which significantly increases the cost of production.

One of the new approaches in LPBF is to use a rescanning strategy of the same layer. The most common application of rescanning is the remelting of the contour and top surface of the manufactured part to decrease surface roughness.[22,23,24] Moreover, the rescanning process might be applied in whole or selected areas of the layer. Even though the multi-scan approach significantly increases a cost of the process, the marked improvement in the quality of the metallic parts is achieved. Two or more scans can be applied for: (1) remelting, which consists of double melting of the layer,[23,25,26,27,28,29,30] (2) presintering and melting[31] as well as (3) melting and scanning with a low energy density to reduce the residual stress.[32] The literature reports beneficial effects of remelting in the LPBF process for various materials. Yasa and Kruth[24] have shown that remelting effectively reduced porosity of LPBF 316L steel, achieved a significant drop in porosity. After three remelting scans, they achieved a significant drop of porosity from 0.77 to 0.032 pct. Additionally, remelting only the top layer has reduced the average roughness (Ra) from 12 to 1.5 µm and contour remelting weakened the staircase effect by about 10-15 pct.[24] Decreased roughness and increased density after remelting in 316L steel was also achieved by Qiu et al.[28] Furthermore, remelting is an effective way to decrease the LOF porosity in LPBF, as shown for 18Ni-300 maraging steel by Song et al.,[27] in which the LOF was reduced from 0.88 to 0.07 pct. In addition, they revealed that after remelting the number of low angle boundaries (LAGB) increased. Moreover, the remelting strategy in LPBF process was applied to pure tungsten[29] and AlSi10Mg alloy[23] also resulted in lower porosity and reduced surface roughness.

Xie et al.[30] investigated the effect of the laser power of the remelting scan in the LPBF in Ti6Al4V alloy. They observed that in the single laser scan the melt pool formation in the powder bed is always asymmetric because of the different thermal conductivity of powder and bulk metal and formation of powder denudation zone. Additionally, interactions with powder particles might be the cause of occurring keyhole porosity, denudation zone, and spatter formation. In turn, during the remelting scan, the geometry of the melt pool is symmetric, which is caused by a more stable process and low spattering, and thus the lower porosity. It was postulated that during the remelting, the laser power in the LPBF process should be higher than in the primary scan to introduce deeper remelting, which results in the reduction of the keyhole porosity.[30]

According to the authors' knowledge, currently, there are no works describing the effect of the remelting energy on the forming microstructure of LPBF Inconel 625. According to our approach, the combination of two scanning strategies with different parameters of the initial and remelting scans, allows to reveal the impact of the energy of remelting scan on porosity and the microstructure. The results obtained in this study are crucial in understanding the influence of remelting energy to enabling the optimization of local microstructure formation of additively manufactured metallic parts of Inconel 625 Ni-based superalloy. The use of remelting scan in LPBF to modify the microstructure and lower porosity of metallic materials holds significant potential for future applications as a method for local or volume modifications. Therefore, investigating the impact of the appropriate process parameters and scanning strategies is crucial to achieving successful implementation.

In light of the challenges and advancements in LPBF, this work aims to investigate the effects of remelting on the properties of LPBF Inconel 625. In addition, we aim to investigate the effect of equal and higher laser power during remelting than in primary scan on the porosity, microstructure and mechanical properties of LPBF Inconel 625. The novelty of this work is the qualitative and quantitative characterization of the effect of remelting on porosity, microstructure, texture, hardness, and tensile strength of LPBF Inconel 625. The parameters used for remelting laser scan allowed the formation of a bimodal microstructure consisting of coarse columnar grains surrounded by narrow bands of smaller grains. Additionally, the use of remelting has a beneficial effect on reducing LOF porosity when these defects are formed due to, for example, the gas flow architecture of the device or a small laser spot, an additional finding from the studies carried out in this work.

2 Materials and Methods

2.1 Characteristics of the Powder

A vacuum inert gas atomized metal powder of Inconel 625 AMPERPRINT® 0153 Ni-SA 625 15-45 (Höganäs, Sweden) was used for the LPBF process. Chemical composition of powder provided by the manufacturer is 21 pct Cr, 9 pct Mo, 4 pct Nb and < 0.01 pct C. Figure 1 presents the scanning electron microscopy (SEM) images of the powder particles used in the LPBF process (Figures 1(a) and (b)). Most of the particles are spherical, but numerous particles with satellites and a splat cap defects are visible. In Figure 1(c) the histogram of powder distribution is presented, based on the performed laser diffraction. The results of the relative volume frequency of the powder particles showed the presence of powder in the range of 8 to 80 µm. The d10, d50 and d90 values were equal to 16, 27 and 45 µm, respectively.

Fig. 1
figure 1

(a, b) SEM images of Inconel 625 powder morphology and (c) size distribution of powder particles

Full size image

2.2 LPBF Process

Inconel 625 samples were fabricated using an AYAS 120 ML machine (Inntec.pl, Poland) equipped with a 200 W single-mode laser with a focus spot diameter equal to 35 µm. The LPBF process was carried out in an argon atmosphere with a maximum oxygen content below 200 ppm. The samples were printed on X2CrNiMo17-12-2 build plates, with 3 mm supports. The initial temperature of the base plate was set at 40 °C. For all samples the remelting scan was performed immediately after the primary laser scan.

The two types of samples of Inconel 625 with different geometry were LPBF manufactured. Cube-shaped samples with the side length of 8 mm were prepared for the microstructure analysis. Moreover, for the tensile test 4 mm × 8 mm × 50 mm dog bone shaped samples with 20 mm gage length were manufactured. The tensile samples were oriented in parallel (∥) and perpendicular (⊥) to the build direction (BD) in order to examine the effect of the microstructure anisotropy on the mechanical properties. In order to obtain the required dimensional accuracy and surface roughness, the tensile samples were grounded and polished.

The samples were prepared in variants without and with additional remelting scan. In all samples, the primary laser scan was performed with a 67° rotation versus the direction of the laser scan in the previous layer. In the remelting samples after the primary scan, additionally a laser scan with bidirectional strategy was performed. The manufacturing strategy scheme is presented in Figure 2. The samples were marked as R0-R3. R0 is the reference sample manufactured only with a single primary scan, while samples R1-R3 are obtained with primary and additional remelting scan strategy. The process parameters are collected in Table I. The process parameters for the first scan were experimentally determined to be within the process window. Relatively high residual porosity level was related to the gas flow depending on the architecture of the device used and the small diameter of the laser spot. The remelting scan parameters were chosen based on the observation of the effect of partial and full remelting of the primary scan melt pools. Most often, the volumetric energy value (VED) is given for comparison of a set of LPBF process parameters, but in the case of a remelting laser scan it is more appropriate to present the energy value in the form of surface energy density (SED), due to the different state of the powder bed and melted metal after the primary laser scan. The SED was chosen to compare the energy delivered during the primary and remelting scan. The SED was calculated using Eq. [1].

$${\text{S}}{\text{ED}}\text{=}\frac{\text{P}}{\text{v*h}}\left[\frac{\text{J}}{{\text{mm}}^{2}}\right]$$
(1)

where P power of the laser, v the scanning speed of the laser beam, and h hatch distance.

Fig. 2
figure 2

Scheme of laser tracks for primary scan and primary and remelting scans in two neighboring layers

Full size image
Table I Parameters of the LPBF Process Used for Manufacturing the Specimens with Single and Remelting Laser Scan
Full size table

2.3 Microstructure Investigation

The microstructure of the samples was investigated by light microscopy (LM) using an Opta-Tech MM100 microscope (Opta-Tech, Poland), as well as SEM using Inspect S50 FEI and Nova NanoSEM 450 microscope (FEI). SEM microstructural investigations were performed using backscatter electron (BSE) and secondary electron (SE) contrast. Electron Backscatter Diffraction (EBSD) was performed with a 20 kV beam voltage and spot size 5 using a FEI Nova NanoSEM 450 SEM microscope equipped with an EDAX EBSD detector. EBSD maps were acquired on an area of 540 × 540 µm2 with the step size of 1 µm for each sample and analyzed using the MTEX[33] toolbox in Matlab 2022a (The MathWorks Inc). Based on EBSD data, inverse pole figures (IPF), the pole figures (PF) and low angle grain boundaries (LAGB) with misorientation angle less than 10 deg and high angle grain boundaries (HAGB) with misorientation angle above 10 deg were determined. The PF were generated for EBSD maps from ⊥BD plane.

Scanning-Transmission electron microscope (STEM) Tecnai TF 20 X-TWIN (FEI) microscope equipped with a STEM high-angle annular dark-field (HAADF) detector was used to microstructure and chemical composition investigations.

For microstructural investigation samples were prepared from ∥ and ⊥ planes. For SEM BSE and EBSD investigations the samples were grounded on SiC grit paper up to P2000 gradation and polished with OP-S suspension (Struers, Denmark). The samples were also electroetched in 60 pct HNO3 for SEM SE investigation. The thin foils for STEM investigations were jet electropolished in a solution of perchloric acid (HClO4) at the temperature − 30 °C.

2.4 Quantitative Microstructural Analysis

The SEM SE and BSE images were analyzed using ImageJ software (Fiji).[34] The equivalent circle diameter of grains (ECD), grain aspect ratio (AR), Feret diameter (Fmax), area of pores, average ECD of the cells, and area fraction of oxide inclusions were measured.

The area fraction (A pct) of pores or oxides were analyzed using the SEM BSE images. The regular net with a fixed mesh size and 1600 points was used for at least eight images for all samples. The total investigated area for measuring of porosity and oxides was at least equal to 2,332,800 and 12,320 µm2, respectively. The nominal magnifications in SEM of 100× and 200× were used for porosity measurements, while for the area fraction of oxides, they were 3000× and 5000×. The area fraction of the oxides were calculated using Eq. [2]. The point method error (δ) was calculated using Eq. [3].

$${A}_{\mathrm{ pct}}=\frac{\sum {P}_{i}*{A}_{i}}{{P}_{T}*{A}_{T}}*100\text{ pct}$$
(2)
$$\delta =\sqrt{\frac{{A}_{\mathrm{ pct}}(1-{A}_{\mathrm{ pct}})}{{A}_{\mathrm{ pct}}*{P}_{T}}}$$
(3)

where: Pi the number of points that hit the feature in the image, PT the total number of points, Ai the area of the image, and AT the total area of all images.

The manual contouring method was used for the EBSD and SEM SE images. It allowed to determine the ECD, Fmax and AR of the grains, as well as the ECD of the cells and pores. The average ECD of the cells was determined using Eq. [4].

$$\overline{ECD }=2\sqrt{\frac{4{A}_{g}}{{N}_{C}*\pi }}$$
(4)

where: Ag the area of grain, Nc number of cells in grain.

The cooling rate (ΔT) for Inconel 625 was calculated only from equiaxed cells using Eq. [5].[35]

$${{\text{ECD}}}_{{\text{cell}}}=134.43*{\Delta T}^{-0.26}$$
(5)

For ECD, Fmax and AR the median values (M), d10, d50, d90 were determined, where d10, d50 and d90 describes the parameter value for the area less than or equal to 10, 50, and 90 pct of the tested area.

2.5 Archimedes Measurements

The porosity of samples was measured by the Archimedes method. The measurements were performed in water at a temperature of about 20 °C using a laboratory scale with ± 0.0001 g accuracy. The average relative density was compared with the density of LPBF Inconel 625 powder equal to 8.47 g/cm3.[15]

2.6 Hardness Measurements

Hardness measurements were performed using the Duramin-40 hardness tester (Struers, Denmark) using a Vickers intender under a load of 1 kg according to ISO 6507 standard. The average hardness was calculated from at least 10 measurements in the middle part of the sample in the ⊥ and a ∥ planes.

2.7 Tensile Tests

Tensile tests were conducted using an Instron 4502 machine (Instron Corp.) with a maximum load capacity of 10 kN at room temperature and a deformation rate of 3 mm/min. The specimens had a gauge length of 20 mm, width of 4.2 mm, and thickness of 2.0 mm. The samples were tested for load direction parallel (LD∥BD) and perpendicular (LD⊥BD) to the BD. The experimental procedure was augmented with the use of a Digital Image Correlation (DIC) System Q400 (Dantec Dynamics A/S, Denmark). This study employed two-dimensional DIC analysis, during the tensile test to acquire the strain distribution. The results were acquired using a high-resolution CCD camera with a 1.9/35 lens. To calibrate the system, a 9 × 9 calibration target with a total side length of 30 mm was used. Before the DIC measurements, the samples were marked by applying a stochastic pattern consisting of a white matte background with black matte speckles by spraying. The evaluations of the strain behavior in the tensile tests were performed using the Istra4D (Dantec Dynamics A/S, Denmark) software.

2.8 Research Methodology

Figure 3 shows the research scheme employed in this study, which consists of several key stages. In the first stage, the Inconel 625 powder is characterized. In the second stage, LPBF Inconel 625 is manufactured with a specific single laser scan and remelting scan parameters. The third stage involves the microstructure characterization of the manufactured samples by LM, SEM, and EBSD techniques. Finally, the mechanical properties of the material are characterized, including hardness measurements and tensile test. Overall, the scheme presented in Figure 3 provides a structured and comprehensive approach to the scope of work for the evaluation of the influence of the remelting laser scan on the microstructure and mechanical properties of LPBF Inconel 625.

Fig. 3
figure 3

Research methodology in this work

Full size image

3 Results and Discussion

3.1 Microstructure

Figure 4 shows the LM images of the microstructure of R0, R1, R2 and R3 samples in the ∥ plane. The microstructure of all samples consists of layers containing crystallized melt pools arranged according to the laser scan strategy. In the R0 sample the melt pools with different widths between layers were observed, which is caused by the 67 deg scanning strategy. The melt pool geometry is affected by process parameters and the depth to width ratio reveals the laser melting mode, which can be conduction, transition or key-hole mode.[36] The geometry of the melt pool in the R0 sample indicates that the LPBF process was conducted in transition mode. In samples R1 and R2, the microstructure consisted of two types of melt pools, the primary melt pools, which are similar to that of R0 and formed as a result of melting the metal powder in the primary scan and shallower secondary melt pools, which were formed during remelting of the previously solidified layer. The penetration depth of the remelting melt pools increases due to the higher laser power. Therefore, in the R3 sample, remelting scan involves large melt pools, while the area of primary scan was limited. The shape of the remelting melt pool contours for the R1 and R2 samples is typical for conduction mode, while for R3 it reveals a transition mode.

Fig. 4
figure 4

Microstructure of (a) R0, (b) R1, (c) R2, (d) R3 samples of LPBF Inconel 625 observed in the plane ∥ BD, LM

Full size image

Different penetration depths and melting modes of primary and remelting laser scanning with the same laser scanning parameters were related to the difference in thermal conductivity of the Inconel 625 powder bed and a bulk material, as well as of the formation of denudation zone and multiple scattering in primary scan. The thermal conductivity of Inconel 625 at 1093 °C in as-printed state and as a powder bed is equal to 26.0 \(\frac{W}{m*K}\) and 0.9 \(\frac{W}{m*K}\) respectively.[37,38] However, the thermal conductivity of the powder bed depends on many factors, such as relative density, powder morphology and pressure of protective gas. Generally, effective thermal conductivity in a powder bed increases with increasing powder diameter and is higher for irregular particles.[39]

The microstructure of Inconel 625 LPBF manufactured samples consists of overlapping solidified melt pools. The overlapping of melt pools occurs between subsequent material layer and laser scans paths. In layer-by-layer remelting, the number density of melt pool boundaries can be greater if the second melt pool is shallower or narrower. When similar energy is applied due to higher thermal conductivity of bulk Inconel 625, a significantly smaller area of material is melted. For that reason, we observed primary and remelting melt pools in R1 and R2 sample, which indicates their formation as a result of remelting of previously solidified layer. Gustmann et al.[26] obtained similar results for partial remelting in the LPBF process of the Cu-11.85Al-3.2Ni-3Mn alloy. They reported that the decrease in depth of remelted melt pools was caused by the application of the higher scanning speed. Our observations of the change of the melt pool mode in remelting from transition to conductive are in line with reports for the AlSi10Mg alloy,[23] and the 18Ni300 maraging steel.[27] The increase of SED of the remelting scan leads to a reduction in the fraction of melt pool boundaries by fully remelting of the primary melt pools. In case of the R3 sample the microstructure formation is mainly affected by parameters of remelting scan.

Based on the melt pool morphology observations the scheme of their formation is presented in Figure 5. In the primary scan, a layer of powder is melted (Figure 5(a)) and a microstructure consisting of melt pools with depth significantly larger than the thickness of the layers was formed (Figure 5(b)). At laser energy in the remelting scan the same or slightly higher than the primary scan (R1 and R2), melt pools shallower than in the primary scan were formed (Figure 5(c)). If the laser power is significantly higher (R3), a similar volume of material might be remelted as in the primary scan (Figure 5(d)). The results show that for Inconel 625 the increase of laser power during the remelting scan should be higher by at least 25 pct to melt a similar volume of material.

Fig. 5
figure 5

Scheme of the melt pool formation (a) primary scan of the powder bed, (b) melt pools geometry after the primary scan, (c) melt pool geometry after remelting with same or slightly higher power (R1, R2), (d) melt pool geometry after remelting with significantly higher power (R3). The dashed black lines indicate the boundaries between subsequent layers, the black solid line the boundaries of the primary melt pools, and the red solid lines the boundaries of the remelting melt pools

Full size image

SEM BSE images of R0-R3 samples microstructure at ∥ and ⊥ planes are given in Figure 6. SEM BSE observations revealed that the sample R0 obtained by a single scan is characterized by a dense structure, but with a noticeable amount of pores. The grains are small and V-shaped with a tendency for elongation along BD in ∥ plane (Figures 6(a) and (b)), while in the ⊥ plane grains are equiaxed in shape (Figures 6(c) and (d)). Such grain morphology is characteristic for Inconel 625 as-printed LPBF samples.[40] The areas between the adjacent laser tracks contain grains of smaller size. Such grains may be formed due to the higher cooling rates in the edge areas of the melt pools.

Fig. 6
figure 6

SEM BSE images of the (a-d) R0, (e-h) R1, (i-l) R2, (m-p) R3 samples of Inconel 625 manufactured by LPBF in the ∥ and ⊥ planes. The dashed yellow lines indicate the boundaries between columnar grains and fine grain bands, while red dashed lines mark the primary melt pool boundaries. Analogically the arrows indicate the direction of laser scanning paths in primary and remelting laser scan (Color figure online)

Full size image

After remelting in the R1 sample, larger and more elongated grains were observed. In the ∥ plane (Figures 6(e) and (f)), the columnar grains were present in the overlapping melt pools of the remelting scan tracks, while small grains were formed bands in the centerline of the melt pool. Meanwhile in the ⊥ plane (Figures 6(g) and (h)), the equiaxed shape of grains was dominant in areas of overlapping melt pools, but in areas of remelting melt pools centerlines smaller grains were also noticed. Due to the lower depth of melt pools during remelting the bottom parts of the primary melt pool are also observed. In these areas, grains are finer with varying shapes.

In the ∥ plane of R2 sample, the microstructure is characterized by grains elongated in BD and small grains bands in the middle of the laser paths (Figures 6(i) and (j)). However, the width of bands of small grains was smaller than in the R1 sample. Similar microstructure was obtained in the R3 sample (Figures 6(m) and (n)), but the columnar grains were wider and number of small grains between columnar grains was very low.

Microstructural analyses of the R1 (Figures 6(g) and (h)) and R2 (Figures 6(k) and (l)) samples in ⊥BD plane revealed that the equiaxed larger grains grown in the areas of overlapped remelting melt pools while the small grains were formed at their centerline. The equiaxed shape was characteristic for cross sections of columnar grains. In the R3 sample (Figures 6(o) and (p)) the large equiaxed grains occupied almost the entire volume and only an insignificant number of small grains was observed in center areas of remelting melt pools. The results of quantitative analysis of grain size and morphology are more detail in the Section III–B.

The results have shown that an increase in the SED of the remelting scan led to an increase in the size of the secondary melt pools, facilitating the epitaxial growth of columnar grains across the melt pools and reducing the fraction of small grain bands. In Inconel 625 subjected to remelting a higher anisotropy of the grain morphology than in a single scan sample was observed. The presence of parallel bands with small grains is observed at a distance of about 60 µm in both planes, which corresponds to the hatch distance. The area fractions of smaller grain bands in the R1, R2, and R3 samples observed in ⊥ plane were equal to 17.3, 14.4, and 6.4 pct, respectively (Figure 6). Our observations of the formation of bands with smaller grains are in good agreement with the results obtained for other materials than Inconel 625 with a bidirectional laser scan strategy.[41,42]

In Figure 7 IPF maps of grain orientation, with marked high-angle grain boundaries (HAGB) (black lines) and LAGB (blue lines) boundaries, as well as the corresponding misorientation angle distribution of the grain boundaries are shown. The graph showing the relative fractions of LAGBs and HAGBs is given in Supplementary Figure S1 (see Electronic Supplementary Figure S1). In the ∥ plane of the R0 sample, the formation of slightly elongated grains is observed. In turn, in the R1-R3 samples subjected to remelting large columnar grains separated by smaller grains bands are present. The grains are elongated in the direction close to the BD. Both in the ∥ and ⊥ planes, the large grains are distributed between bands of smaller grains that are present in the middle of the melt pools. With increasing SED of remelting, a widening of the columnar grains and reduction of the area of smaller grains bands occurred.

Fig. 7
figure 7

IPF maps with superimposed grain boundaries and low angle boundaries and corresponding distribution of grain boundaries misorientation angle for (a,b) R0, (c,d) R1, (e,f) R2, and (g,h) R3 in the ∥ and ⊥ planes. HAGB > 1 deg are marked in black, while LAGB < 10 deg by blue) (Color figure online)

Full size image

The fraction of LAGBs in the R0 sample is about 17 pct, while for the R1-R3 samples the corresponding values are at least twice higher. For the R0 sample, the high relative amount of HAGBs was observed due to 67 deg laser scan rotation during the process. Similar values of grain boundaries misorientation than in R0 were observed by Sun et al,[43] Kreitcberg et al. in LPBF Inconel 625 obtained with 67 deg scan rotation. In case of bidirectional scanning strategy the amount of LAGBs in LPBF Inconel 625 is almost twice higher than for 67 deg scan direction, which was presented by Fang et al.[40]

In turn, in the remelted samples, we observed an increase in the number of LAGBs due to the increase in the relative amount of fine grain bands and the formation of subgrain boundaries in the coarse columnar grains. Even though the relative amount of LAGBs is similar in samples R1-R3, in the R1 and R2 samples the many of them are located in fine grain bands due to large residual thermal strain.[44] Additionally, in the R2 and R3 samples, LAGBs are formed as subgrain boundaries inside columnar coarse grains, which was confirmed by the increase of fraction of 2 to 5 deg LAGBs. Additionally, columnar grains with similar orientation provide the formation of high density of LAGBs between grains distributed in adjacent melt pools, as a direct result of competitive growth and temperature gradient.[45]

4 Quantitative Microstructure Analysis

The results of the grain size and shape analysis performed with the use of EBSD IPF maps are collected in Table II. The quantitative analysis show that the R0 sample exhibits the smallest grains from all investigated samples with ECD up to 88 µm. In the ∥ plane approximately 61.1 pct of area fraction of the grains are classified as columnar, while in the ⊥ plane, grains are smaller and more equiaxed with fraction of elongated grains equal to 39.2 pct. The anisotropy of the microstructure is the result of the epitaxial grain growth during the LPBF process, migration of solid-liquid interfaces and side-branching of grains.[46] It can be observed that almost half of the analyzed area of the sample consists of grains that are longer than twice the thickness of the layer. Formation of such a microstructure is associated with a 67 deg rotation of the laser paths in subsequent layers, causing disruption of epitaxial grain growth and significant refinement of the microstructure.[42]

Table II ECD, Fmax, AR and Area Fraction of Columnar Grains for the ∥ and ⊥ Planes of Examined Inconel 625 Samples
Full size table

In the R1-R3 samples subjected to remelting significant changes in the distribution of ECD, Fmax, and AR occurred. The tendency of formation of columnar grains elongated in BD with a high fraction of grains longer than several hundred micrometers were noticed. For the R1 sample the trend of formation columnar grains, with maximum Fmax of about 330 µm is visible. Meanwhile, more than 50 pct of the area fraction of the R2 and R3 samples is covered by a large grains with Fmax larger than 395 and 528 µm, respectively. Moreover, in the R1-R3 samples we observed an increase of the AR and Fmax of small grains. It indicates that grains in small grains bands are also elongated in BD.

In the ⊥ plane of R1 and R2 samples, most of grains are with the Fmax less than 60 µm, which is equal to the length of hatching distance. However, in the R3 the increased number of grains with Fmax greater than 60 µm corresponds to the merging of grains separated by LAGBs from adjacent laser paths (Figure 7).

The columnar microstructure was formed after remelting with a 180 deg laser scan rotation in successive layers. There coarse and elongated columnar grains in the middle of the laser paths and fine columnar grains in the areas between the laser paths were formed. Most of the fine grains between the laser paths observed on ⊥ plane had the ECD smaller than 15 µm. The area fraction of such small grains decreased with increasing of remelting SED, and for the R1, R2, and R3 samples and were 11.4, 9.0, and 3.7 pct, respectively.

The elongated grains in the ∥ plane are common in additively manufactured Inconel 625, but their size and AR depend on the parameters of the LPBF process. The microstructure in as-printed state is mostly dependent on the SED and laser scanning strategy. The scanning path rotation significantly influences the texture and grain morphology in LPBF manufactured metals. Leicht et al.,[42] in 316L steel observed the formation of large columnar grains with {011}〈001〉∥ BD orientation, which were separated by narrow bands of {001}〈001〉∥ BD oriented columnar grains in the centerline of melt pools at the bidirectional scanning strategy. After applying a laser scan rotation by 90 and 45 deg the microstructure became more complex without formation of bands with columnar grains. The 67 deg rotation provides the lowest anisotropy of the microstructure.[42] On the other side, Wan et al.[47] observed in the LPBF manufactured Inconel 718 a bimodal structure, due to applying a bidirectional scanning. In our study for the bidirectional remelting laser strategy, we observed the formation of a columnar microstructure similar to that achieved after the single bidirectional rotation strategy in the 316L steel[42] or in the Inconel 718.[47] However, due to mixing of strategies in primary and remelting scans, as well as formation of shallower melt pools in R1 and R2 the microstructure is significantly affected by the primary scan.

4.1 Grain Orientation and Texture

PFs determined based on the EBSD results are presented in Figures 8 and Supplementary S2 (see Electronic Supplementary Figure S2). The weakest texture was obtained in the R0 sample with minimum and maximum texture coefficients for the (001) plane equal to 0.45 and 2.4. In turn, the strongest texture occurred in the R3 sample with minimum and maximum texture coefficients equal to 0 and 9.4. For the R1 and R2 the maximum values were lower than for the R3 sample, however they suggest the formation of the strong texture. To indicate the texture components, the ideal pole locations of Goss ({110} 〈001〉) and Cube ({001} 〈001〉) were superimposed on the (001) PFs.

Fig. 8
figure 8

(001) PFs with superimposed theoretical texture components of Cube and Goss for all grains, grains with ECD > 15 µm and ECD < 15 µm for (a) R0 (b) R1, (c) R2, and (d) R3. The BD, transverse direction (TD) and scanning direction (SD), as well as, maximum and minimum texture coefficient are marked

Full size image

The PF of the R0 sample shows a very weak texture without dominant components. In the R1 and R2 samples the locations of texture maxima are close to the ideal position of Cube and Goss components. Meanwhile, for the R3 sample the texture maxima are located in good agreement with ideal Goss locations. Using the MTEX software, the PFs for the R1-R3 samples were obtained for the grains with ECD larger and smaller than 15 µm (Figure 7). The grains with ECD smaller than 15 µm were located mostly in smaller grain bands, while those with ECD larger than 15 µm were identified as coarse columnar grains. In the R1 sample for coarser grains the maxima of texture components were located close to the Goss ideal position, whereas for finer grains, they were close to Cube. Strong Goss texture components were also identified in the R2 and R3 samples in columnar grains areas. In finer grains regions, the stronger Cube component was observed. For the R1 and R2 samples the texture coefficients for both types of grains were high. For the R3 sample the values of texture coefficients were much smaller for finer grains, but it might be related to the low area fraction of smaller grains in this sample.

In additively manufactured Ni-base superalloys, such as Inconel 625 and Inconel 718, the literature reports that the dominant texture is Goss[40,48] or Cube,[47,48,49] but they can change according to the process parameters. Ma et al.[48] present the hypothesis that in Inconel 625 additively manufactured by the laser directed energy deposition process, texture formation is mainly affected by the direction of heat flow, which in turn depends on laser power and beam diameter, as well as laser scan strategy.

In FCC crystals during solidification, the primary grain growth direction is to 〈001〉 crystallographic direction. In the LPBF process, epitaxial growth of grains is only possible when the grains in the previous layer are oriented in 〈001〉 crystallographic direction and aligned parallel to the local heat flow direction.[50]

The formation of Cube and Goss texture in different parts of the melt pools are strongly related to heat flux. The {001}〈001〉∥ BD oriented grains were in centerlines as effect an epitaxial grain growth with very high cooling rates according to the parallel heat flux in BD, while the coarse columnar grains were aligned about 45 deg to the BD providing the {110}〈011〉∥BD texture. The strong texture along the BD is often observed in AM nickel-based superalloys, but it highly depends on the LPBF process parameters. Our results showing the formation of bimodal grain size microstructure and texture are in good agreement with previous studies of LPBF Inconel 718[41] and 316L steel[46,51] manufactured with bidirectional scanning strategy.

In the R1-R3 samples in the middle of laser paths the direction of heat flow is parallel to the BD, which provides the formation of Cube texture fine grains. The coarser columnar grains formed in areas where the overlapping of the melt pools occurs and the direction of heat flow significantly deviated from the BD, which provide to nucleation of new Goss-oriented grains as an effect of side-branching of solidification cells in adjacent laser paths.[46,51]

The fraction of Cube oriented grains decreases due to the increase of the SED of remelting scan due to changing the melt pool mode from conductive to transition and providing the greater melt pool overlapping. Analogically, Nadammal et al. observed a similar effect due to increases in higher hatch spacing in LPBF Inconel 718,[41] while Liu et al.,[52] on the other side, observed a decrease in the intensity of texture with increasing scanning speed, due to shallowing of the melt pool in LPBF Inconel 718. The formation of a strong crystalline texture is influenced by various process parameters such, as the efficiency in removing gas vapors and spatters during the LPBF process by inert gas flow. This observation was made by Andreu et al.[53] in their investigation of LPBF 316L steel. They observed a stronger crystalline texture when scanning directions facilitated easier removal of metal vapors leading to higher energy irradiating the surface.[53] Due to the higher heat conductivity of bulk metal than powder bed, during remelting there is no intensive formation of spatters and metal vapors, which improves the energy irradiation efficiency and formation of strong texture along BD in Inconel 625.

The formation of a strong {011}〈001〉∥BD texture in Inconel 625 depends not only on the scanning strategy used in LPBF, but also on the size, shape and overlap of the melt pools. More stable melting process in remelting can be used for microstructure tailoring due to the possibility of using process parameters outside the primary process window.

Our results confirm that microstructure formed after remelting is more dependent on the parameters of the remelting scan than on parameters used during primary scan, but the depth and shape of the melt pools have major impact on the grain growth and texture formation. Due to increasing SED of remelting scan the transition from a fine grains microstructure with low texture to a bimodal grain size with strong texture occurred. Although a microstructure consisting of only large columnar grains was not obtained, the use of remelting can be considered as a potential method to produce single-crystal-like microstructure in Inconel 625 Ni-base superalloy which belongs to the conventionally wrought polycrystalline superalloys. The use of additional remelting laser scan without feeding a new layer of powder might be used to multiply the effect of growth of previously formed grains.

4.2 Porosity

The pores in the R0-R3 samples were observed on SEM BSE images (Figure 9). The R0 sample exhibited the highest porosity, while the R3 sample was almost fully dense. The porosity measured by Archimedes method was 0.33 ± 0.02, 0.25 ± 0.02, 0.01 ± 0.02 and 0.02 ± 0.02 pct for R0, R1, R2 and R3 samples, respectively. The maximum ECD of observed pores was equal to 98, 71, 52 and 6 µm for R0, R1, R2 and R3 samples, respectively. In all samples, some amount of gas porosity was observed. In addition, in the R0, R1 and R2 samples, the LOF porosity was present due to the incorporation of spatters from the LPBF process. This type of porosity was characterized by irregular pores, which were located under particles a few times larger than the virgin Inconel 625 powder. During the laser melting process around the keyhole inlet, some metal droplets are ejected in the direction opposite to the laser scan.[10,54] Most of the metal droplets should be removed by gas flow, but some of them fell into the powder bed and were incorporated into the material in the form of spatters. The spatters are characterized by a larger size and higher oxidation compared to virgin powder, making them more difficult to melt it in the LPBF process.[55] Due to the increase in SED in the R2 and R3, the laser beam was sufficient to melt spatters during the remelting scan, which is the main cause of the decrease in porosity.

Fig. 9
figure 9

Porosity of the (a) R0, (b) R1, (c) R2 and (d) R3 samples of Inconel 625. The types of pores observed are marked by arrows

Full size image

Our results are in good agreement with literature reports for other materials subjected to remelting in LPBF process. Yu et al. observed a decrease in LOF porosity after remelting in 17-4 PH steel.[56] Similarly, Song et al. under remelting conditions obtained a significant decrease in the number of large pores in 18Ni-300 maraging steel.[27]

Porosity has a great influence on the mechanical properties of LPBF-manufactured metallic parts. Increased porosity could deteriorate their tensile strength, ductility, and fatigue properties.[57] The decrease in the number of the LOF pores after remelting of Inconel 625 obtained in our study should have a positive effect on mechanical properties by reducing the number of areas of stress concentration. This type of processing might be an excellent alternative to HIP.

4.3 Chemical Composition and Cellular Microstructure

Conventionally manufactured Inconel 625 is a solid solution-strengthened alloy, but during rapid solidification in LPBF process, a dendritic or cellular-dendritic microstructure, with small precipitates in the intercellular areas is formed. In as-printed Inconel 625 the precipitated phases were identified as Nb- and Mo-rich MC carbides, as well as Mo, Cr, Fe, and Si-rich Laves phase.[58,59,60] In SEM BSE (Figure 6) and SE (Figure 10 and Electronic Supplementary Figure S3) the cellular microstructure inside the grains is visible with cells with a ECD smaller than 0.6 µm. In each unique grain the cell size and shape is related to the direction of grain growth. In the intercellular areas we observed presence of submicrometric precipitates (Figure and Electronic Supplementary Figure S3). Based on the EDS measurements it was revealed that they are enriched in Nb and Mo and depleted in Ni and Cr (see Electronic Supplementary Figures S3 and S4). We observed a small depletion of the Cr and Ni concentration in the intercellular areas as well as in the precipitates. Simultaneously the precipitates were enriched with Nb, which is consistent with Lass et al.[61] An observed increase in the intensity of the Nb Lα and Mo Lα peaks suggests the possible presence of Nb-rich carbides or Laves phases in the cell walls. STEM-HAADF images of single scan (R0) and remelted (R3) samples (Figure 11) revealed that such precipitates, which are nanometric in size, are located mainly in the areas of the cell walls. A high density of tangled dislocations was also observed. Oxide particles, spherical or hexagonal, are visible in dark contrast in SEM BSE (Figure 13) and STEM-HAADF (Figure 11) images. The fine oxides are evenly dispersed in cells and cell walls.

Fig. 10
figure 10

Low and high magnification SEM SE images of cellular microstructure of (a, b) R0, (c, d) R1, (e, f) R2, (g, h) R3 samples of Inconel 625 LPBF at the ⊥ plane. The dashed yellow lines indicate the boundaries between columnar grains and fine grain bands, while red dashed lines indicate the primary melt pool boundaries (Color figure online)

Full size image
Fig. 11
figure 11

STEM-HAADF image of (a) R0 and (b, c) R3 samples and (d) EDS spectra and chemical composition at point P1 and area A1 marked in (c)

Full size image

Characterization of submicrometric precipitates in Inconel 625 LPBF was the subject of previous studies.[15,60,62] In this work, we did not observed a significant influence of remelting scan strategy on the formation of secondary phases.

In Figures 10, 11 and Supplementary S3 (see Electronic Supplementary Figure S3) the differences in cell size inside the melt pools can be observed, while in Figure 12 the distribution of cells diameter and cooling rate were presented. The finest cells occur in areas with higher cooling rates, such as melt pool boundaries, whereas in the middle and upper parts of melt pools the solidification cells were coarser because of lower cooling rates. In all investigated samples the grains in areas characterized by the highest cooling rate contain of cells with ECD of about 0.15 µm. In samples R1-R3 subjected to remelting, two groups of grains were observed. First, those with a greater cells with average ECD between 0.20 to 0.35 µm for R1 and 0.25 to 0.40 µm for R2 and R3, which were located in the coarse grains, and grains with finer cells located on the edge of the remelting melt pools and finer grain bands. The average ECD of cells in the small grain bands in the R1, R2 and R3 samples was in the range of 0.15 to 0.20 and 0.20 to 0.25 µm, respectively. Our observation of the different cell sizes according to the position in of the melt pools is consistent with the results of Gan et al.[38], who observed in the Inconel 625 the finest cells in the bottom and near the periphery areas of the melt pools and the largest in the top and middle parts. Microstructural analysis showed that for the R0, R1 and R2 samples, similar solidification cell size was observed in the areas of the overlapping melt pools. Finer cells were present in the peripheral parts of the melt pools in the remelted samples. This may be due to the different cooling rates in the powder bed and solidified Inconel 625. In turn, in the R3 sample, we observed a small increase in the ECD of the cells compared to R0, probably due to the higher laser power used in the remelting scan. No significant increase in cell size suggests that either no significant heat accumulation occurs during the primary scan or it is balanced by the higher thermal conductivity of the solidified Inconel 625.

Fig. 12
figure 12

Distribution of cooling rate values estimated based on cell diameter measurements in (a) R0, (b) R1, (c) R2, (d) R3 samples

Full size image

4.4 Oxide Inclusions

The LPBF process is prone to the formation of dispersed oxide particles. The oxygen content in AM metals is 4 to 6 times higher than that of conventionally manufactured materials.[63] Oxide particles, spherical or hexagonal, are visible in dark contrast in SEM BSE (Figure 13) and STEM-HAADF (Figure 11) images. The fine oxides are evenly dispersed in cells and cell walls. Nanometric oxide particles, identified as Al2O3, are common in Inconel 625 manufactured with LPBF and other Ni-based superalloys.[60,62,64]

Fig. 13
figure 13

SEM BSE images showing the distribution of oxide particles in (a) R0, (b) R1, (c) R2, (d) R3 samples

Full size image

To examine the influence of remelting on the amount of oxide particles, their area fraction was determined. The results are given in Figure 13. The average area fraction of oxide inclusions for R0 and R1 is almost the same. It was noticed that with an increase in laser power in remelting scan the area fraction of oxides also increased. The diameters of oxides particles were less than 100 nm, and they were homogeneously dispersed throughout the samples without a tendency to segregation.

The sources of oxygen in the LPBF process are the contamination of the powder at the stage of inert gas atomization, moisture pollution, and extraneous oxygen in the protective gas atmosphere during the manufacturing process.[63] During the process, trace oxygen from the gas shielding atmosphere interacts directly with the melt pool and can also be introduced with oxidized metal spatters.[55,63] In the remelting scan strategy, the interaction time of shielding gas with the melt pool is at least twice longer than in a single scan, which might enhance the oxidation of the metal. Moreover, in R2 and R3 samples, a decrease in spatter-related porosity occurred, which could be related to an increase in the number of oxides. Accordingly, no significant increase in the oxide area fraction in R1 sample might be correlated with a large number of nonmelted spatters.

It is well-known that the cooling rate significantly affects the size and number of oxides in AM metals. Yu et al. observed relevant differences in the oxide content, its size and morphology in Inconel 718 LPBF additively manufactured by electron beam melted (EBM) and LPBF. They observed the formation of small nanometric oxides during the LPBF, whereas in the EBM the oxide inclusions were a few times larger. The main differences between these two methods are the lower cooling rates and the increase in the powder bed temperature in EBM.[65] Also, Deng et al. observed a larger size and a higher area fraction of oxides in gas-atomized stainless steel powders than after the LPBF process.[63] This indicates that they dissolute during the melting by a laser beam and are formed again during cooling. Therefore, the observed increase in oxide area fraction in R2 and R3 samples might be an additionally enhanced by a slightly lower cooling rate.

The homogeneous distribution of nanosized oxide dispersoids in LPBF Inconel 625 should have a strengthening effect by dislocation bowing around the oxide particles according to Orowan mechanism.[66] Further results of the investigation of the role of oxide dispersoids in Inconel 625 LPBF are given in the Sections III–G and III–H.

4.5 Hardness

The results of the hardness measurement in samples in ∥ and ⊥ versus BD are presented in Supplementary Figure S5 (see Electronic Supplementary Figure S5) and Table III. Hardness values of LPBF Inconel 625 was higher than that of wrought Inconel 625 by approximately 40 to 70 HV[67,68] and are in the similar range as reported by others for as-printed Inconel 625.[15,69] Li et al. obtained the initial hardness of Inconel 625 equal to 343 HV,[69] while Marchese obtained 293 HBW (308 HV) for the lowest porosity sample.[15] The hardness of the R1-R3 samples was higher than of the R0 sample. Furthermore, for the R0 and R1 higher hardness was in the ⊥ plane, while for the R2 sample a higher hardness value was obtained for the ∥ plane. In turn, for R3 sample no significant differences in hardness were observed in both analyzed planes.

Table III Parameters Characterizing Mechanical Properties of the Test Samples Obtained in this Work Compared with the Data for Materials Examined by Others
Full size table

In AM materials, the strengthening effect is related to very high cooling rate and the formation of small grains, cellular-dendritic structure, high density of dislocations, fine precipitates and oxide dispersoids.[66]

The hardness is synergistically affected by many microstructure factors, so it is difficult to determine which is dominant. According to the Hall–Petch relation, the hardness is inversely proportional to the square root of the mean grain size. In our work, we did not observe a decrease in hardness for samples after remelting, for which a significant increase in grain size was achieved. This may suggest that there is compensation for hardness through other mechanisms. A natural explanation seems to be that the effect of the cell walls is similar to that of the grain boundaries. However, there is currently no consensus on the mechanism of strengthening from the cellular microstructure. Hong et al.[70] observed that cells with a high density of dislocations in LPBF 316L steel act as grains boundaries.

On the contrary, Wang et al.[71] showed that in the AlSi10Mg alloy an increase in strength is only due to an increase of the density of dislocations located mainly in the cell walls. Despite the discrepancy between the results, it can be assumed that the small cell size promotes the strengthening of the material due to the high dislocation density along cell walls. In R1 and R2 samples, we observed the formation of multiple melt pool boundaries after remelting, and consequently a higher volume of areas with fine cells than in R0 sample. The formation of LAGB, as well as fine cells, could be the reason of increased resistance to dislocation gliding, and thus a higher hardness and YS.

Oxide inclusions can also have a significant impact on hardness. According to the Orowan mechanism, the hardness should increase with increasing oxide fraction and decrease of their size. An almost threefold increase in the area fraction of oxide particles achieved in the R2 and R3 samples compared to the R0 and R1 samples would also contribute to the increase in the average hardness of these samples due to dispersion strengthening. The strengthening effect is similar to that of oxide-dispersion strengthened alloys.[45] Incoherent oxide inclusions act as efficient pinning points for dislocations.[72] Smith et al. observed the strengthening effect of oxide dispersion in Laser Direct Energy Deposited manufactured austenitic steels, but due to their larger size, it was much smaller than in conventional ODS alloys.[66]

Another aspect regarding microstructure-hardness relationship is the difference in hardness depending on the plane of measurement. For R0 and R1 samples, the hardness was higher in ⊥ plane, which is most likely due to the lower ECD of the grains and as a result the strengthening effect corresponding to the Hall–Petch effect.

For the R2 sample, the higher hardness occurred in the ∥ plane, despite of significantly larger ECD of the grains. In this case, the effect from multiple overlapped melt pools and the presence of small grain bands might provide the higher hardness. In turn, in the R3 sample, no significant differences in hardness occurred in both planes, which corresponds to a reduction of areas of small grain bands and a significant increase of grains size compared to the R2 sample.

4.6 Mechanical Properties

The results of the uniaxial tensile test combined with the DIC analysis are shown in Figure 14. The samples were tested in the loading direction parallel (LD∥BD) and perpendicular (LD⊥BD) to the BD. The size, shape, and orientation of grains is similar to those in the base of the tensile samples (see Electronic Supplementary Figure S6). In Table III the results of mechanical properties for investigated samples, as well as, the literature reports were collected.

Fig. 14
figure 14

(a,f) Engineering stress - strain curves determined in tensile tests of R0, R1,R2 and R3 samples with LD∥BD and LD⊥BD and DIC images of tensile strength in LD∥BD and LD⊥BD before rupture, and true strain distribution maps determined from DIC analysis for (b, g) R0, (c, h) R1, (d, i) R2 and (e, j) R3

Full size image

The results showed that the R0 sample had a similar YS and the elongation for both loading directions was approximately 30 pct. However, it was characterized by about 100 MPa higher tensile strength for LD⊥BD.

For the R1-R3 samples, an anisotropy of properties related to ductility was observed, which was associated with an anisotropy of grain morphology. In the case of LD∥BD, a higher elongation was observed, which was above 50 pct, whereas in the case of LD⊥BD, the elongation value was lower or close to R0 sample.

The YS values for R1 and R2 were slightly higher, and for R3 slightly lower than for R0. This is most likely related to the bimodal grain distribution, and the strengthening occurred due to the interaction of fine grain bands with coarse columnar grains. In the case of sample R3, the tensile response was mostly related to the large columnar grains.

The obtained values of YS, UTS are definitely higher than the values given in the standards for the conventionally produced Inconel 625 alloy. The minimum required YS, UTS, and elongation for the annealed Inconel 625 wrought solution according to ASTM B443 are equal to 276 MPa, 690 MPa, and 30 pct, respectively. Meanwhile YS and UTS for the Inconel 625 LPBF according to ASTM F3056-14 are equal to 275 and 485 MPa, respectively.[58] The mechanical properties of the R0 sample were very similar to the results presented in other works for LPBF Inconel 625 in as-build (AB) and stress-relieved (SR) conditions.[18,58,68,73,74]

For the samples tensile tested parralel to BD, YS, UTS, and elongation are higher than the minimum required values for wrought and LPBF Inconel 625. Only the R1 and R2 samples tested with LD⊥BD exhibited elongation lower than 30 pct according to the strong grain morphology anisotropy. The very high elongation for remelted samples in LD∥BD and UTS values are similar to those for samples produced with solution annealing[73] or HIP,[21,75] but with a higher YS about 110 to 220 MPa. The higher YS is connected to the presence of tangled dislocations in the area of cell walls, which provide resistance to dislocations gliding.

All of the samples showed anisotropy of the tensile properties, depending on the LD versus BD orientation. This type of anisotropy is characteristic for materials manufactured by the LPBF process. A higher UTS and lower plasticity in the LD⊥BD are generally observed. This is related to the texture and grain morphology anisotropy, and the melt pool orientation.[73,76]

The DIC analysis confirm that the plastic properties of specimens tested with LD∥BD are characterized by significantly higher values in comparison to LD⊥BD. The sample R0 tensile tested in LD∥BD showed multiple areas of the strain localization at the last step before cracking (Figure 14(b)). In this case, the crack propagation at an angle of 45 deg was observed. The numerous areas of strain localization were observed in the gauge length, which may be related to the presence of LOF porosity and nonmelted spatters combined with lower grain morphology anisotropy than in remelted samples. In the case of the R1, R2, and R3 samples the strain localization was concentrated in the middle area of the gauge length (Figures 14(b) through (d)). The highest elongation was obtained for the R2 sample, which is clearly shown on the DIC strain distribution maps (Figure 14(d)) as well as on the stress - strain tensile curves (Figure 14(a)). The decrease of porosity as well as the formation of columnar grains, might have a significant effect on more uniform strain distribution.

In the case of the samples tensile tested with LD⊥BD the strain values were lower than for LD∥BD. The observed character of the crack propagation was different. In all samples, the crack at an angle of 45 deg was observed (Figures 14(g) through (j)). The lowest strain value was observed for the R1 sample, and the highest value was recorded for the R3 sample. In all samples with LD⊥BD the multiple strain localization was observed. For the R3 sample, the strain distribution is more homogeneous, which might be related to the lowest porosity, as well as by low area fraction of small grain bands (Figure 14(j)).

Both the DIC analysis and the stress–strain tensile curves showed that the strain for the samples with LD⊥BD was almost two times lower than for samples tested with the LD parallel to BD (Figure 14), which corresponds to the anisotropy of grain morphology of the analyzed samples.

Despite the presence of very large columnar grains, the contribution of small grains, which are mainly located in bands between them, is significant and leads to the formation of a large number of HAGBs. This type of microstructure can lead to interesting properties in which large grains promote increased ductility, while small grains corresponding with the strength of Inconel 625 LPBF. It is similar to findings obtained by Li et al.[77] for CoCrWNi alloys with a bimodal grain distribution or by Flipon et al.[78] in spark-plasma sintered 316L steel, which demonstrated an increase in the YS due to the increase in the volume fraction of small grains. Accordingly, the decrease in YS for R3 compared to other remelted samples could be related to the lowest volume fraction of small grains.

LPBF Inconel 625 is very often subjected to additional heat treatment consisting of annealing and solution annealing and aging. Marchese et al.[58] observed significant changes in mechanical properties after heat treatment in the temperature range of 1080 °C and 1150 °C, which resulted in the recrystallization. Annealing at temperatures and the formation of equiaxial grains with annealing twins, which make the structure similar to in wrought materials.[58] In contrast to remelting after recrystallisation, a significant decrease in hardness, YS, and UTS associated with an increase in ductility was reported.[58]

A significant advantage of remelting over annealing leading to recrystallization is the increase in the plasticity of the samples tested with the LD∥BD with increased hardness, YS and UTS. However, there is a slight deterioration of plasticity in LD⊥BD. These results suggest that remelting has potential application in LPBF manufacturing of nickel-based superalloys characterized by low porosity, highly anisotropic columnar grain microstructure and with a large amount of LAGBs. Additionally, it can be used to locally change the microstructure for optimal properties.

5 Conclusions

Laser beam remelting in the LPBF process is a powerful method to reduce a porosity, but also tailoring the microstructure of Ni-based superalloys and other metal materials. This study presents the effect of SED, applied by adjusting the laser power during remelting laser scan, on the formation of a bimodal microstructure.

The study compared the influence of remelting at the same and higher SED compared to the primary laser scan. The results showed that at similar SED, the remelted area in the subsequent scan was smaller than in the initial scan. To achieve similar sizes and shapes of remelted melt pools increasing in energy by approximately 25 pct is necessary.

The different shapes and sizes of the remelted melt pools have a significant impact on the resulting microstructure and crystallographic texture. When the melting depth during remelting is smaller than in the initial laser scan, the alloy microstructure and properties are mainly determined by the initial scanning strategy. Particularly, in samples with lower SED, the formation of a bimodal microstructure was observed, consisting of columnar grains with Goss texture and high fraction of fine grain bands with Cube texture distributed in the melt pools centerlines. In the case of the highest investigated SED, the remelted melt pool areas were large enough that the influence of the primary scan was negligible, and the resulting microstructure consisted only of large columnar grains with a high fraction of LAGB and limited area of smaller grain bands.

An additional effect of remelting was a decrease in porosity and an increase in the volumetric presence of nanoscale oxides. During the subsequent remelting process, it was observed that a higher energy is required to melt the spatters incorporated on the surface and the to reduce LOF porosity that forms during the primary melt. The use of remelting has a beneficial effect on reducing LOF porosity, which results of spatter incorporation caused by gas flow architecture of the device or a small laser spot. Furthermore, the small increase in the area fraction of nanoscale oxides is partly due to the remelting of oxidized spatters. The applied remelting laser scan in LPBF for Inconel 625 resulted in the formation of a strong microstructural anisotropy, which also led to mechanical anisotropy. A significant increase in ductility was observed in the direction of columnar grains after remelting, with a slight decrease in UTS. Additionally, the formation of a bimodal microstructure contributed to an increase in YS as the effect of the primary scan.

Remelting in LPBF can be used for both increasing the reliability of materials by reducing the porosity and as a method for controlled tailoring of the microstructure and, thus, the mechanical properties.

The results obtained in this study demonstrate that applying laser beam remelting can create a more complex microstructure compared to a single scan, which can lead to variations in strength parameters. Furthermore, remelting can be a promising technique for manufacturing Ni-based superalloys with a single-crystal-like structure. Moreover, remelting is a good technique for reducing the limitations of process control in printers with relatively small laser beam diameters and without the possibility of gas flow control. However, remelting as a modification of the LPBF process carries several risks, such as increases in the build rate, and changes the thermal history of the elements, which might have an effect on the material properties. An increase in contact time between the liquid metal and the atmosphere during the process, is a reason of increased area fraction of oxides.