Microstructure Evolution in Homogenization Heat Treatment of Inconel 718 Manufactured by Laser Powder Bed Fusion

Abstract

This study systematically investigates the homogenization-induced Laves phase dissolution kinetics and recrystallization mechanisms in laser powder bed fusion (L-PBF) processed IN718 superalloy. The as-built material exhibits a characteristic fine dendritic microstructure with interdendritic Laves phase segregation and high dislocation density, featuring directional sub-grain boundaries aligned with the build direction. Laves phase dissolution demonstrates dual-stage kinetics: initial rapid dissolution (0–15 min) governed by bulk atomic diffusion, followed by interface reaction-controlled deceleration (15–60 min) after 1 h at 1150 °C. Complete dissolution of the Laves phase is achieved after 3.7 h at 1150 °C. Recrystallization initiates preferentially at serrated grain boundaries through boundary bulging mechanisms, driven by localized orientation gradients and stored energy differentials. Grain growth kinetics obey a fourth-power time dependence, confirming Ostwald ripening-controlled boundary migration via grain boundary diffusion. Such a study is expected to be helpful in understanding the microstructural development of L-PBF-built IN718 under heat treatments.

1. Introduction

Inconel 718 (IN718), a nickel-based superalloy, derives its exceptional mechanical properties from coherent γ″-Ni₃Nb precipitates with body-centered tetragonal DO₂₂ structure and γ′-Ni₃(Al,Ti) phases with cubic L1₂ structure within its γ matrix [1,2]. These precipitation-hardening mechanisms endow IN718 with superior creep resistance, high-temperature strength, and corrosion stability under extreme thermo-mechanical conditions, making it indispensable for critical applications in aerospace, marine, and nuclear industries [3,4,5]. However, conventional manufacturing techniques impose significant limitations on geometric complexity, prompting interest in laser powder bed fusion (L-PBF) additive manufacturing. This layer-wise fabrication method enables production of intricate geometries through localized energy deposition guided by digital CAD models [4,6,7].

The L-PBF process subjects materials to rapid thermal cycling characterized by localized melting, directional solidification, and repeated thermal transients. This results in two primary metallurgical challenges: (1) substantial residual stress accumulation from steep thermal gradients, and (2) microsegregation-induced formation of non-equilibrium phases, including deleterious Laves phases and carbides [8,9]. Numerous studies have documented the detrimental effects of these microstructural features. Sui et al. [10] demonstrated that elongated Laves phases act as stress concentrators and primary void nucleation sites during room-temperature tensile loading. Yoshinori et al. [11] further identified these intermetallic compounds as preferential pathways for crack initiation and propagation under cyclic loading conditions. Residual stresses compound these issues by modifying effective stress ratios during fatigue loading [12,13] and influencing fracture toughness through local stress state alterations [14].

Post-process heat treatments represent a critical pathway for microstructural optimization in L-PBF IN718. Unlike conventional cast IN718, which exhibits coarse dendritic structures and severe macro-segregation, L-PBF IN718 features ultrafine cellular-dendritic grains (1–10 μm) with localized microsegregation due to rapid solidification rates (>10⁶ K/s). For instance, cast IN718 typically requires extended homogenization times (>20 h) to dissolve Laves phases, whereas L-PBF IN718 achieves similar dissolution within 1–3 h due to refined microsegregation. Current research emphasizes the temperature-dependent dissolution kinetics of Laves phases and stress-relief mechanisms. Li et al. [15] reported incomplete Laves phase dissolution following 980 °C treatments, while homogenization at 1065 °C for 1.5 h achieved significant microsegregation reduction. Contrastingly, Tucho et al. [16] observed persistent precipitates even after 1250 °C/7 h treatments, suggesting complex dissolution dynamics. Importantly, extended high-temperature exposures risk excessive grain coarsening while eliminating residual stresses, creating an optimization challenge between precipitate dissolution and microstructural refinement [4,17,18,19].

The current understanding of homogenization kinetics primarily derives from conventional processing routes. For cast IN718, Javad et al. [20] established Laves phase dissolution kinetics at 1150 °C, while Miao et al. [21] developed time–temperature homogenization models, demonstrating morphology-dependent dissolution behavior. However, the distinct microstructural architecture of L-PBF materials—featuring ultrafine grains, cellular solidification structures, and unique solute segregation patterns—renders conventional heat treatment protocols inadequate. This technological gap is particularly evident when comparing with forged IN718 systems where grain growth kinetics during recrystallization have been extensively characterized for thermomechanical processing optimization [22].

This study addresses critical knowledge gaps by systematically investigating Laves phase dissolution kinetics and grain growth behavior in L-PBF IN718 during homogenization treatments. Through quantitative microstructural analysis and kinetic modeling, we aim to establish process–structure relationships that enable the rational design of heat treatment protocols. The developed kinetics framework will provide predictive capabilities for microstructural evolution while balancing competing requirements of precipitate dissolution, stress relief, and grain size control, which are essential for achieving optimal mechanical performance in additively manufactured superalloy components.

In this work, we focus specifically on the differences in homogenization response and microstructural evolution of Inconel 718 processed under various solidification conditions. The effects of subsequent thermomechanical treatments and other post-processing routes are not discussed in this study, as our aim is to first establish a clear understanding of how initial solidification microstructures influence homogenization behavior. Further research will be required to systematically investigate the combined impact of initial structure and complex post-processing on the final properties of Inconel 718 components.

2. Material Preparation

Fabrication was conducted using a Renishaw AM 250 system equipped with a 200-W SPI Yb-fiber laser (Renishaw, Gloucestershire, UK). Key process parameters are detailed in

Table 1. The parameters of the L-PBF process.

Laser Power (W)	Point Distance (μm)	Exposure Time (μs)
200	90	100
Layer Thickness(μm)	Hatch Spacing(μm)	Laser Spot Diameter(μm)
30	90	70

. A C45E4 carbon steel substrate (100 × 60 × 10 mm³) was mechanically polished (SiC abrasive to 1200 grit) and degreased with acetone prior to installation. Gas-atomized IN718 powder (AMC Powders, 15–45 μm diameter distribution) with spherical morphology (Figure 1a) and chemical composition verified by ICP-OES (

Table 2. The chemical compositions of IN718 powder.

Element	Ni	Cr	Nb	Mo	Ti	Al	C	Fe
Content (wt.%)	54.28	17.97	5.33	3	0.98	0.57	0.024	Bal.

) was utilized. Powder batches were dehydrated at 120 °C for 3 hr in a vacuum oven (<50 Pa) to minimize the moisture content. The build platform was maintained at 80 °C throughout processing to mitigate thermal stress gradients. A rotational stripe scanning strategy (67° interlayer rotation) was implemented to minimize directional anisotropy (Figure 1b). A numerical model was developed for the powder melting and solidification process in LPBF. The computational fluid dynamics (CFD) method employed in this model is the highly computationally efficient Lattice Boltzmann Method (LBM).

Homogenization treatments were conducted in a high-purity argon atmosphere (99.999%) using a KSL 1600X tube furnace (Hefei Kejing Material Technology Co., Ltd., Hefei, China). As-built specimens underwent isothermal annealing at 1150 °C with dwell times spanning 2–90 min (2, 5, and 10-min increments from 10–90 min), followed by immediate water quenching to preserve high-temperature microstructures.

Nanoindentation mapping was performed on orthogonal planes (Z-Y and X-Y orientations) using a Berkovich diamond tip. Vickers microhardness testing was conducted with an MHV-1000BZ automated system (Shanghai Jiezhun Instrument Equipment Co., Ltd., Shanghai, China) under 500 gf load with 15 s dwell time. Triplicate measurements were acquired per condition. All specimens were ground up to 5 μm grit size and polished by a metallographic polisher and vibrating polishing machine. To reveal the grain and precipitates structures, these specimens were etched in a solution consisting of 150 mL H₂O + 250 mL HCl + 2.5 g CrO₃. The microstructures of all specimens were observed by a Keyence VHX-2000 optical microscope (OM) (Osaka, Japan) and TESCAN VEGA IILMH scanning electron microscope (SEM) (Brno, Czech Republic) equipped with an electron backscatter diffraction (EBSD) unit and an energy dispersive spectrometer (EDS). The step size of the EBSD measurements was 2 μm. The image processing software Image-Pro-Plus 6.0 was used for measurements. Ten figures were used for measuring the volume fractions of the Laves phases.

3. Results

3.1. Microstructure of L-PBF-Built IN718

The EBSD microstructure and crystallographic characterization of as-built IN718 are shown in Figure 2. Figure 2a shows that the grains have obvious elongation along the building direction, which is a characteristic of L-PBF and has been observed by many researchers [23,24,25]. Most grains are filled with red, which shows the as-built IN718 with a <100> preferred crystallographic directions. As shown in the grain size distribution histogram (Figure 2b), the microstructure exhibits heterogeneous grain sizes ranging from 2.25 μm to 10 μm, yielding an average grain size of 8.63 μm. The average aspect ratio of 2.27 further confirms the anisotropic grain morphology induced by the layer-wise deposition process. Crystallographic texture analysis (Figure 2c) reveals a weak cubic texture component ({100} <100>), evidenced by the intensity maxima in the {100} pole figure. This texture development aligns with the competitive growth of <100>-oriented dendrites during rapid solidification. This strong crystallographic texture induces significant anisotropy in mechanical properties. The as-built IN718 with large residual stresses represents the high local stresses and high dislocation densities [26]. The approach to calculating geometrically necessary dislocations (GND) densities is based on the strain gradient model [27]. The GND density $\rho$ is related to the misorientation angle $\mathsf{\vartheta }$, and it can be described by $\rho =2\vartheta /\mu b$, where $\mu$ is the unit length and $b$ is the magnitude of the Burgers vector. The value of local misorientations can be retrieved directly from the EBSD data of kernel average misorientation (KAM). Figure 2d shows the GND density distribution in the yellow rectangular area of Figure 2a. The values of GND density vary from about 2.1 × 10¹² m⁻² to around 1.5 × 10¹³ m⁻² in the grain interior. It should be noted that the 2 μm EBSD step size limits the spatial resolution for detecting the finest substructures and may lead to an underestimation of the local GND density, especially when using KAM-based calculations. Figure 2e shows the distribution of the grain boundary in an individual grain, and the boundaries of zones with a misorientation of between 2°~5°, 5°~15°, and >15° are drawn in red, green, and black, respectively. The high density of low angle grain boundaries (LAGBs) correlates with the observed GND concentrations and exhibits directional alignment parallel to the build axis. This sub-grain structure originates from dislocation network formation during cyclic thermal stressing in the L-PBF process, creating a hierarchical microstructure with <100>-oriented columnar grains containing fine cellular substructures.

Figure 3 presents cross-sectional microstructural features and corresponding nanoindentation responses across orthogonal planes. Optical microscopy (OM) analysis in Figure 3a,d reveals effective interlayer metallurgical bonding and continuous columnar growth spanning multiple deposition layers (white arrows indicate the build direction). High-magnification SEM observations demonstrate distinct solidification morphologies: the Z-Y plane exhibits fine cellular dendritic structures with preferential alignment along thermal gradients (primary dendritic arm spacing, λ₁ = 600–900 nm), while the X-Y plane displays quasi-equiaxed cellular morphology, representing characteristic transverse sections of dendrites. Energy-dispersive spectroscopy (EDS) confirms that the irregular bright striated phases (indicated by arrows in Figure 3b,e) correspond to Nb-rich Laves intermetallic compounds, predominantly concentrated in interdendritic regions, consistent with microsegregation patterns in L-PBF-processed Ni-based superalloys.

Nanoindentation mapping reveals significant anisotropic mechanical behavior, as evidenced by the load-displacement curves in Figure 3c,f. The Z-Y plane shows an average reduced modulus of 135.97 GPa (σ = ±8.4 GPa), whereas the X-Y plane demonstrates a markedly higher value of 214.14 GPa (σ = ±11.2 GPa). This modulus discrepancy (58.3% increase in the X-Y orientation) and high data dispersion (coefficient of variation of 16–18%) originate from two key factors: (1) crystallographic anisotropy, where <100>-oriented grains exhibit lower stiffness along the build axis, and (2) heterogeneous phase distribution, manifested by contrasting mechanical responses between Laves phase-enriched regions (Vickers hardness ~650 HV) and γ-matrix domains (~350 HV). Nanoindentation tests indicate that the elastic modulus and hardness in the X-Y plane are notably higher than those measured in the Z-Y plane. This discrepancy is attributed to the preferential orientation of <100>-textured grains and the alignment of cellular substructures and dislocation networks, which locally reinforce the transverse direction. The mechanical response anisotropy revealed here can be directly linked to the solidification-driven microstructural features originating from the L-PBF process. This pronounced anisotropy has important implications for component performance and post-processing, particularly in machining and forming operations where uniform material response is desirable.

Figure 4 presents high-magnification SEM micrographs illustrating the microstructural characteristics of Z-Y and X-Y planes in the as-built IN718 sample. As shown in Figure 4a, the dendritic growth direction deviates approximately 15–25° from the deposition direction, with statistical analysis revealing a primary dendrite arm spacing (PDAS) of 0.80 ± 0.12 μm. Notably, numerous irregular bright phases exhibiting elongated filamentary morphology (indicated by arrows) are preferentially distributed within interdendritic regions, exhibiting characteristic topological alignment along interdendritic channels. EDS compositional mapping (Figure 4c,d) confirms that these secondary phases are enriched in Nb and Mo compared to the γ-matrix, unequivocally identifying them as Laves phases formed through microsegregation-induced elemental partitioning.

3.2. Microstructure of Homogenization-Treated IN718

Figure 5 systematically tracks the dissolution kinetics of Laves phases under isothermal homogenization at 1150 °C with time-dependent morphological evolution. As demonstrated in Figure 5a,b, compared to the as-built condition featuring elongated Laves precipitates, a 2-min treatment initiates significant phase fragmentation, resulting in granular morphologies with reduced aspect ratios and refined equivalent diameters Prolonged holding to 30 min (Figure 5d,e) drives progressive spheroidization and size reduction, accompanied by topological isolation of residual Laves particles.

Figure 6 presents inverse pole figure (IPF) maps and grain size distributions (excluding annealing twin boundaries) to elucidate microstructure evolution during homogenization at 1150 °C. After 2 min treatment (Figure 6a), epitaxial columnar grains persist with strong crystallographic alignment, maintaining the directional solidification signature inherited from L-PBF. A notable microstructural transition occurs at 5 min (Figure 6b), where strain-induced recrystallization initiates, evidenced by the emergence of equiaxed grains. As seen in Figure 6c–f, it is obvious that grain size increases gradually with the increase of homogenization-time.

Figure 7 delineates the recrystallization dynamics and grain boundary evolution of IN718 during homogenization at 1150 °C. Grain boundary bulging and some new recrystallized grains can be found at 1150 °C for 2 min, as shown in Figure 7a. After full recrystallization, grain boundaries are pinned by some second phase particles, as shown in Figure 7b. Figure 7c shows the statistical results of recrystallization at different holding times. The reported percentages of recrystallized and substructure grains are calculated from pixel-by-pixel EBSD data with area confidence indices above 97%, indicating high dataset reliability. The measurement uncertainty is thus mainly attributable to data analysis and interpretation rather than conventional band detection methods. It can be seen that nearly 100% of the area in the as-built IN718 is labeled as deformed grains. The proportion of deformed grains decreased slightly, and the proportion of recrystallized grains increased slightly when the samples were held for two minutes. When the holding time increased to 5 min, the deformed grains were basically eliminated, and the ratio of recrystallized grains and substructure grains was 47.5% and 50.3%, respectively. As the holding time continues to increase, the substructure grains proportion gradually decreases, and the recrystallization proportion gradually becomes the main grain composition. The variation of recrystallization proportion agrees with the change of LAGBs proportion at different holding times. The proportion of LAGBs in the L-PBF-built IN718 is as high as 0.65, as shown in Figure 7d. After 2 min, the ratio of LAGBs slightly decreases, which is consistent with the presence of new grains in Figure 7a. With the increasing of the holding time, the proportion of LAGBs in the range of 2 to 15° is significantly reduced, and the number of HAGBs is also obviously increased. Additionally, the ratio of twin boundaries increases gradually until the homogenization time exceeds 30 min, the proportion of LAGBs in the range of 2° to 15° is very small, and only a small amount of LAGBs less than 2° exists.

After homogenization, the significant reduction in <100> fiber texture intensity and the grain morphology change from columnar to equiaxed collectively reduce the mechanical anisotropy of the alloy. The as-built material exhibited notable differences in hardness and elastic modulus between the X-Y and Z-Y planes, attributable to the crystallographic alignment and microstructural features induced by the L-PBF process. However, heat treatment promotes recrystallization and randomization of grain orientations, leading to more uniform mechanical performance across different directions. This microstructural evolution is highly desirable for engineering applications requiring consistent and predictable material behavior under multi-axial loading conditions.

4. Discussion

4.1. Microstructure of L-PBF-Built IN718

Figure 8a,b demonstrates that the weak {100} <100> cubic texture and refined grain structure in L-PBF-processed IN718 originate from the complex interplay between solidification dynamics and thermal history. The observed dendritic morphology exhibits distinct spatiotemporal variations governed by thermal gradients (G) and solidification rates (R) within the molten pool. Dendritic growth orientation deviations of 10–25° from ideal <100> directions arise through competitive growth between crystallographically equivalent <100> orientations ([100], [010], [001]), driven by three synergistic factors: melt pool geometry-induced thermal gradient vector variations, epitaxial growth from polycrystalline substrate seeds with diverse orientations, and 67° interlayer rotation modifying local heat flow patterns [28].

As illustrated in the solidification mode diagram (Figure 8c), the quantified process parameters (G: 3.7 × 10⁶–2.8 × 10⁷ K/m, R: 0.2–0.8 m/s) confirm cellular-dendritic growth dominance, aligning with predictions from the Kurz–Fisher solidification model. Figure 8d reveals significant thermal gradient variations across melt pool regions, with high G values of 2.8 × 10⁷ K/m at the solidification front (purple profile) decreasing to 3.7 × 10⁶ K/m at the trailing edge (red profile). Primary dendrite arm spacing (PDAS) calculations using the modified Kurz–Fisher equation yield 0.6–1.5 μm spacing ranges, consistent with experimental measurements in Figure 3. The ultra-high cooling rates (>10⁶ K/s) induce substantial undercooling, promoting grain refinement through three interconnected mechanisms: constitutional supercooling-driven nucleation events [29], competitive growth elimination of unfavorably oriented dendrites, and thermal stress-induced sub-grain formation (2–5° misorientation).

The formation of Laves phases results from the micro-segregation of various elements, which have lower solute distribution coefficients (k < 1), such as Nb and Mo in the solidification of L-PBF. During the solidification process of IN718, the solute distribution coefficient of Mo, Nb is lower than 1. These elements will segregate to the liquid in inter-dendritic regions, and the segregation of Nb has an important influence on the formation of Laves phase [30]. In casting IN718 alloy, the Nb content of Laves phase is in the range of 22.3–30 wt.% and the blocky Laves phase is observed in the center of the inter-dendritic region, as reported in previous research works [31,32]. The higher cooling rate of L-PBF process can eliminate the degree of macro-segregation, but it is difficult to avoid micro-segregation. The segregation of Nb is a typical phenomenon due to solute rejection and redistribution during the solidification process.

4.2. Dissolution Behavior of the Laves Phase

Figure 9a quantifies the temporal evolution of the Laves phase volume fraction during homogenization of L-PBF IN718 across temperatures ranging from 1100 °C to 1200 °C, revealing distinct dissolution kinetics governed by the empirical relationship:

$${\phi }_{Laves}={\phi }_0\mathrm{e}\mathrm{x}\mathrm{p}(-K_T\ast \mathrm{t})$$

(1)

where ${\phi }_0$ is the initial Laves phase volume fraction of L-PBF-built IN718, t is the homogenization time, $K_T$ is the parameter depending on homogenization temperature. The measured ${\phi }_0$ by metallographic area analysis is 16.85 ± 2.83% in the present experiment. Contrary to conventional expectations [20], complete dissolution exhibits non-monotonic temperature dependence, achieving full elimination at 1150 °C within 3.7 h—an order-of-magnitude faster than cast IN718 (27.7 h at 1150 °C [33]). These accelerated kinetics arise from L-PBF’s characteristic microstructural advantages: reduced macro-segregation and refined dendritic morphology. The dissolution timeframe follows the generalized model [34]:

$$\tau =Bexp(-0.036T)$$

(2)

where τ is the time to dissolve the Laves phase completely, T is the homogenization temperature; and B is decided by the segregation degree. It has been reported by Javadd [20] that the content of the Laves phase cannot always decrease with the increase of homogenization temperature, and the content almost does not change significantly with the extension of time. Therefore, it is assumed that the dissolution of the Laves phase is complete when the solubility of the Laves phase is greater than 80%. Based on the time to dissolve the Laves phase completely of 3.7 h at 1150 °C, parameter B was determined as 3.53 × 10¹⁸. Miao et al. [21] and Javad et al. [20] have calculated the parameter as 2.33 × 10¹⁹ and 2.64 × 10¹⁹, respectively, which is obviously higher than the calculated value in the present work. Compared to the casting IN718, the L-PBF-IN718 has less segregation degrees and smaller primary dendrite spacing, so it has a better element diffusion ability and smaller parameter B than the casting IN718. When the Laves phase completely dissolves, Equation (1) can be described by:

$$3.37\%=16.85\%exp(-K_T\ast t)$$

(3)

By performing the natural logarithm on Equation (3), $K_T$ is obtained by Equation (4):

$$K_T=-{\displaystyle \frac{\mathit{ln}\frac{3.37}{16.85}}{t}}$$

(4)

The time $t$ of dissolving the Laves phase completely can be expressed by Equation (2). Substituting Equation (4) into Equation (1), the model of Laves phase dissolution kinetic can be deduced, as shown in Equation (5):

$${\phi }_{Laves}=16.85\mathrm{\%}exp\left[(-4.56\times {10}^{-19}t)/exp\left(-0.036T\right)\right]\times 100\mathrm{\%}$$

(5)

Based on Equation (5), the dissolution process of the Laves phase at homogenization temperatures of 1100 °C, 1125 °C, 1175 °C, and 1120 °C can be quantitatively predicted, as shown in Figure 9a. It is clear that the dissolution rate of Laves phase is significantly different in different stages. With the increasing of holding time, the dissolution rate of laves phase gradually slows down. The Laves phase dissolution process mainly includes the subsequent steps: decomposition of the Laves phase, the transfer of atoms through the interface towards the matrix and the diffusion of these atoms into the matrix [35]. To describe this particle dissolution, two physical models have been developed, including the effects of long-distance diffusion [33] and non-equilibrium conditions at the interface [36]. In the following, the dissolution kinetic models of atomic diffusion and interfacial reaction are used to determine the main control factors of Laves phase dissolution in homogenization heat treatment of L-PBF-built IN718.

The dissolution kinetics of the Laves phase can be modelled by Johnson–Mehl–Avrami–Kolmogorov (JMAK) based on element diffusion [37], as shown in Equation (6):

$$X=1-exp (-k{t}^n)$$

(6)

where X is the relative change of the volume fraction of the Laves phase $\phi$ ($X=∆\phi /{\phi }_0$), t is the time, and the exponent $n$ depends on the material properties. The constant k is dependent on the temperature, and it obeys the Arrhenius-type relationship:

$$k=k_{0}exp \left(-{\displaystyle \frac{Q}{RT}}\right)$$

(7)

where $k_0$ is a constant, $R$ is a gas constant, $Q$ is the activation energy, and $T$ is the temperature. For calculation of $k$ at each homogenization temperature, the Equation (6) can be described by:

$$\mathrm{l}\mathrm{n} (-\mathit{ln}\left(1-X\right)=\ln K+n\ln t$$

(8)

The variations of the Laves phase volume fraction with homogenization time are shown in Figure 9a. The plots of $\mathrm{l}\mathrm{n}(-\mathit{ln}\left(1-X\right)$ − $\ln t$ are shown in Figure 9b. The value of the intercept is $\ln K$ at different homogenization temperatures. Based on Equation (7), the activation energy for Laves phase dissolution can be calculated by:

$$\ln k=\mathit{ln}{k}_0-{\displaystyle \frac{Q}{RT}}$$

(9)

As a result, the absolute value of $Q$ was determined as ~605 kJ/mol based on the slope in Figure 9c. Chlebus et al. [35] found that the dissolution of Laves phase is concerned with the diffusion of Nb. This result is not consistent with the diffusion activation energy of Nb (438.7 kJ/mol) reported in the literature [38], which indicates that element diffusion is not the only mechanism of dissolution kinetics in the whole dissolution process of the Laves phase. The higher apparent activation energy for Laves phase dissolution in L-PBF IN718 may arise not only from atomic diffusion processes but also from interface curvature effects, solute segregation at the phase boundary, and localized chemical inhomogeneity, all of which can impede interface migration and raise the energy barrier compared to bulk Nb diffusion.

The interface reaction has a strong effect on the dissolution kinetics for spherical particles by the changing of the solute concentration at the particle–matrix interface [39]. Assuming that the dissolution kinetic of the Laves phase is controlled by the interface reaction, the kinetic model of spherical particles can be expressed by means of the following general law [40]:

$$3\left[1-{(1-X)}^{{\displaystyle \frac{1}{3}}}\right]=k_{r}t$$

(10)

where $k_r$ is a measure of the rate of the interface reactions relative to the rate of long-distance diffusion. Figure 9d demonstrates the shift from diffusion-controlled to interface-controlled kinetics during Laves phase dissolution at elevated temperatures. When t ≥ 10 min, the theoretical and experimental values have a good linear correlation. This indicates that the interface reaction plays an important role in the dissolution process of the Laves phase.

The observed non-monotonic temperature dependence reflects the transition from bulk diffusion-limited kinetics to interface-controlled kinetics at later stages. At lower temperatures, atomic mobility is rate-limiting, resulting in strong temperature dependence. At higher temperatures or at later stages, the dissolution rate plateaus as interface reactions (e.g., boundary migration, segregation effects) become dominant, reducing the overall sensitivity to temperature. Above all, the dissolution of the Laves phase is controlled by atomic long-distance diffusion in the early stage of homogenization. With the increasing of holding time, the Laves phase was further dissolved, and the chemical potential gradient between the particle and matrix gradually decreases. So, the interface reaction rate gradually slows down. The dissolution of the Laves phase is mainly controlled by the interface reaction process in the later stage of Laves phase dissolution. It should be emphasized that cast IN718 exhibits strong macrosegregation, resulting in large chemical potential gradients and favoring rapid interface-controlled Laves phase dissolution. In contrast, L-PBF IN718 exhibits finer microsegregation, which limits chemical gradients and requires both diffusion and interface-controlled mechanisms for complete dissolution.

4.3. Recrystallization Behavior

The as-fabricated L-PBF IN718 alloy exhibits pronounced residual stresses that induce localized plastic deformation during manufacturing, generating substantial variations in geometrically necessary dislocation (GND) densities between adjacent grains. Quantitative analysis reveals GND density differences, creating stored energy gradients that drive grain boundary serration when the driving force exceeds the capillary resistance. This results in characteristic serrated boundaries, which serve as preferential sites for recrystallization nucleation through thermal grooving and boundary bulging mechanisms. However, the mobility of these boundaries is significantly constrained by Zener pinning effects from intergranular Laves phases and MC carbides.

Detailed analysis of misorientation evolution during homogenization demonstrates distinct recrystallization stages. In the as-built condition (Figure 10a,d,g), cumulative point-to-origin misorientation exceeds 10° along 60–75% of analyzed paths. After 2 min at 1150 °C (Figure 10b,e,h), rapid sub-grain coalescence reduces maximum misorientation, accompanied by a decrease in LAGB density. Prolonged homogenization for 5 min (Figure 10c,f,i) promotes twin boundary formation while residual misorientations fall below 1°, indicating recrystallization completion. Comparative analysis reveals fundamental differences in recrystallization behavior between L-PBF and conventionally processed IN718. The L-PBF variant demonstrates a significantly higher recrystallization onset temperature (1100 °C) compared to forged IN718 (980–1020 °C) and directionally solidified counterparts (1150–1180 °C). These delayed kinetics arise from distinct nucleation mechanisms: sub-grain merging dominates in L-PBF IN718, contrasting with the grain boundary nucleation preference observed in forged material.

Upon completion of recrystallization driven by stored energy relaxation, the microstructure remains metastable, with fully recrystallized grains exhibiting continued growth at elevated temperatures through prolonged annealing, as demonstrated in Figure 11a. Precise control of post-heat treatment microstructures and properties requires fundamental understanding of grain evolution kinetics under varying thermal conditions. The temporal dependence of grain growth can be mathematically described by the classical kinetic relationship [41]:

$$G^n-{G_0}^n=k\times t$$

(11)

where $G_0$ is the grain size of IN718 at 1150 °C for 5 min, $G$ is the average grain size at homogenization time $t$, $k$ is a temperature-dependent parameter, and $n$ is the grain growth exponent. The exponent $n$ depends on the grain growth mechanism. When the value of n is 2, 3, or 4, it represents a pure system with no defects or precipitates, precipitate phases with diffusion in the grain, and precipitate phases with diffusion in the grain boundary, respectively. Additionally, the value of $n$ is also affected by element segregation, and the size and distribution of the second phase in the alloys [42]. By performing the natural logarithm on differentiated Equation (11), Equation (12) is obtained.

$$\mathit{ln}\left({\displaystyle \frac{dG}{dt}}\right)=\mathit{ln}\left({\displaystyle \frac{k}{n}}\right)+\left(1-n\right)\mathit{ln}G$$

(12)

In Equation (12), ${\displaystyle \frac{dG}{dt}}$ is calculated as below:

$${\displaystyle \frac{d{G}_n}{d{t}_n}}={\displaystyle \frac{G_n-G_{n-1}}{t_n-t_{n-1}}}$$

(13)

The n and K values at 1150 °C depend on the linear regression of the plot of $\ln \left({\displaystyle \frac{dG}{dt}}\right)$ against $\mathrm{l}\mathrm{n}G$, as shown in Figure 11b. A fixed n value of 4 indicates the Ostwald ripening of the precipitates by grain boundary diffusion [40,42], and K is calculated as 2.85 × 10¹¹ at 1150 °C. It was not suitable to describe the experimental data with other n values, such as 2 or 3, as can be seen in Figure 11. The experimentally observed grain growth exponent in this study (n ≈ 4) is significantly higher than the classical value (n ≈ 2) typically anticipated for systems controlled by pure grain boundary diffusion. This deviation indicates that additional factors, beyond grain boundary diffusion, play a substantial role in governing grain growth kinetics during homogenization. One such factor is the pronounced particle pinning effect exerted by residual MC-type carbides, which remain stable and undissolved at the homogenization temperature of 1150 °C. As evidenced by Figure 11b, these carbides are located at grain boundaries and serve as persistent obstacles to grain boundary migration. This phenomenon, referred to as Zener pinning, significantly retards grain boundary movement, thereby restricting grain coarsening despite prolonged exposure to elevated temperatures. Consequently, the average grain size remains relatively fine, and the grain growth kinetics are suppressed compared to those of systems lacking particle pinning. These findings demonstrate that the presence of undissolved second-phase particles plays a critical role in stabilizing the microstructure and elevating the apparent grain growth exponent during homogenization heat treatment.

Compared with the grain growth of forging-IN718 controlled by the presence of δ phase particles at the grain boundary, Laves phases and carbides existing at grain boundaries tend to hinder boundary motion in L-PBF-IN718. Additionally, L-PBF-IN718 has a large recrystallized grain size and uneven distribution, which is slightly larger than the grain size of the forging-IN718 alloy of about 20 μm.

4.4. Vickers Hardness Test

The homogenization process significantly alters mechanical properties through concurrent stress relaxation and dislocation annihilation. The Zener–Wert–Avrami formalism effectively describes thermal stress relaxation kinetics [43]:

$${\displaystyle \frac{{\sigma }^{RS}}{{{\sigma }_0}^{RS}}}=exp\left[{-\left(A{t}_a\right)}^m\right]$$

(14)

where ${{\sigma }_0}^{RS}$ is the original residual stress before heat treatment, ${\sigma }^{RS}$ is the ultima residual stress at temperature $T_a$ for $t_a$ time, $m$ is the numerical parameter depending on the dominant relaxation mechanism, and A is expressed by:

$$A=Bexp\left[-{\displaystyle \frac{∆H}{k{T}_a}}\right]$$

(15)

where $B$ is a constant, $k$ is the Boltzmann constant, and $∆H$ is the activation enthalpy for the relaxation process. For laser shock peened In718, Zhou [44] determined the associated parameters $m$ = 0.77, $∆H=3.45 \mathrm{e}$ V, and $B$ = 2.18 × 10¹² min⁻¹. The residual stress relaxed and the hardness after each stage of heat treatment can be estimated as in Figure 12. The heat treatment at 1150 °C for 5 min caused a reduction in hardness of around 36%, reaching a stable hardness about 200 HV. Meanwhile, the residual stress is removed about 80% at 1150 °C for 5 min, and after 10 min, the residual stress is completely removed (Figure 12b). Therefore, the residual stress has an important role in strengthening the microstructure. Both mechanisms—residual stress relief and grain coarsening—play a part in hardness reduction. However, the residual stress relaxation afforded by heat treatment occurs much more rapidly than significant grain growth, thus dominating the initial decrease in hardness in this work.

It should be noted that this study was limited to evaluating the effects of different solidification conditions on microstructural evolution during homogenization heat treatment. The influence of further thermomechanical processing and other post-treatment stages on microstructure and properties were beyond the scope of the present work. Future investigations are needed to explore how various post-processing strategies, in conjunction with initial solidification microstructures, affect the final performance of the alloy. Such studies will be essential for optimizing manufacturing routes for additively manufactured Inconel 718.

5. Conclusions

In the present study, the laser-based powder bed fusion of metals was employed to manufacture the IN718. The effects of the homogenization parameters on Laves phase dissolution kinetic and recrystallization behavior of IN718 have been investigated. The main findings are summarized as follows:

(1)

The microstructure of L-PBF-built IN718 has a fine dendritic structure, with a large amount of Laves phases precipitating in the inter-dendritic region. The matrix has a high dislocation density, and the sub-grain boundaries have a certain directivity.

(2)

The dissolution of the Laves phase is controlled by atomic long-term diffusion in the early stage of homogenization. With the increasing holding time, the interface reaction speed slows down, and the dissolution of the Laves phase is mainly controlled by the interface reaction process. The equation of Laves phase dissolution can be described as $R_{Laves}=exp\left[(-4.56\times {10}^{-19}t)/exp\left(-0.036T\right)\right]\times 100\mathrm{\%}$.

(3)

High dislocation density or local-orientation at the serrated GBs are a large driving force through grain boundary bulging for the recrystallization nucleation. The pinning effect of the Laves phases and carbides makes the mechanism of sub-grain merging nucleation dominant. The growth of newly formed sub-grains along the deposition direction may be related to the directivity of sub-grain boundaries.

(4)

Grain growth is mainly controlled by the pinning of the second phase at grain boundaries. The exponent $n$ of 4 that determines the mechanism of grain growth explains that the Ostwald ripening of the second phase particles through grain boundary diffusion during homogenization treatment.