The Role of Temperature Field Distribution in the Microstructural Evolution of High-Strength Aluminum Alloys During Laser Powder Bed Fusion

Abstract

Laser powder bed fusion (LPBF) of high-strength aluminum alloy 7075 (AA7075) is severely limited by hot cracking. However, the underlying mechanisms, particularly the coupling between thermal fields, solidification microstructure, and cracking behavior, remain insufficiently clarified. This study elucidates these mechanisms by integrating experimental characterization with thermal simulation to investigate the temperature field, microstructure, and cracking relationships in both AA7075 and a crack-resistant 7075-Er-Zr alloy. Results show that coarse hot crack morphology is highly dependent on linear energy density E_L. In AA7075, E_L < 450 J/m promotes laterally inclined cracks (short, narrow cracks extending from the melt pool boundary toward the track center), whereas E_L higher than that value leads to the continuous centerline cracks (long, wide cracks along the track center). Fine microcracks are also observed at melt pool boundaries. The 7075-Er-Zr alloy demonstrates superior crack resistance. At E_L = 600 J/m, longitudinal centerline cracks still penetrate along the track, but the alloy achieves crack-free tracks at 200 W with scanning speeds above 1000 mm/s, otherwise exhibiting only short discontinuous cracks. Microcracks at melt pool boundaries are markedly suppressed in the modified alloy. The enhanced crack resistance is attributed to Er/Zr-induced grain refinement and a transition to an equiaxed grain structure, which disrupts intergranular gaps. Critically, thermal simulations identify an annular region with a peak temperature gradient. In AA7075, this region develops aligned columnar grains that facilitate both microcracks and centerline cracks. In the 7075-Er-Zr alloy, microcracks are fully eliminated within this region. However, a residual crystallographic texture persists in the annular region, which promotes the continued occurrence of centerline cracks under high energy density (e.g., E_L = 600 J/m). The annular region remains a critical weak link, and its microstructural control determines the prevailing crack type. This work provides a fundamental understanding of the thermal-microstructural origins of cracking and offers a theoretical foundation for developing crack-resistant aluminum alloys via LPBF.

1. Introduction

Laser powder bed fusion (LPBF) is an additive manufacturing technique that enables one-step, near-net-shape fabrication of complex geometries, offering significant advantages in lightweight engineering fields such as aerospace [1,2]. However, high-strength aluminum alloys (e.g., aluminum alloy 7075, AA7075) are highly susceptible to hot cracking during LPBF, which severely limits their practical application [3,4]. The high cracking susceptibility of the AA7075 alloy arises from its intrinsic properties, including a wide solidification range [5], a large solidification shrinkage (≈6.6%) [6], and limited eutectic liquid films [7]. In addition, the extremely high temperature gradient (G) and solidification velocity (R) within the melt pool promote the epitaxial growth of coarse columnar grains [8]. These columnar grains tend to align their grain boundaries parallel to the build direction, creating long, continuous liquid-film channels that facilitate strain localization and hinder liquid feeding through the mushy zone. According to the RDG model [9], crack susceptibility is related to the pressure drop in the mushy zone and liquid feeding capability.

Modulating the solidification microstructure to promote the columnar-to-equiaxed transition (CET) is an effective strategy to suppress hot cracking [10,11]. According to the criterion, CET occurs for:

$${\displaystyle \frac{G^n}{R}}<Cst$$

(1)

With

$$Cst=a{\left\{\sqrt[3]{{\displaystyle \frac{-4\pi N_0}{3\ln \left(1-\phi \right)}}}\right.·\left.{\displaystyle \frac{1}{n+1}}\right\}}^n$$

(2)

where φ is the prediction probability of columnar morphology (0 being fully equiaxed and 1 being fully columnar), N₀ is nucleation density of the alloy, and n and a are material-dependent constants. Reducing the temperature gradient or increasing the solidification velocity favors the formation of equiaxed grains. An alternative and more practically feasible approach in LPBF is to introduce heterogeneous nucleation sites through alloy modification. Elements such as Sc, Zr, Ti, and Er have been added to aluminum alloys to generate in situ Al₃X particles [12,13,14,15], which lower the nucleation barrier and induce equiaxed grains even at relatively high Gⁿ/R values. This breaks the continuous growth of columnar grains, relieves stress, and thereby inhibits cracking. For instance, Al₃Sc has been recognized as the most potent nucleant for α-Al due to its low lattice mismatch (≈1.32%) and high thermal stability. Al₃Zr provides coherent nucleation sites with a lattice mismatch of about 0.75%, effectively refining grains under LPBF conditions [16,17]. Al₃Ti can exist as a metastable L1₂ phase under the high cooling rates of LPBF, which also promotes α-Al nucleation [18]. More recently, Er has drawn attention as a potential grain refiner [19]. However, their application in LPBF often requires additional powder homogenization [20,21], which is costly.

It is noted that even in the presence of additional nucleation particles, inappropriate processing conditions can still lead to crack formation [22,23]. This indicates that the mesoscale distribution of the thermal field, i.e., the spatial variation in temperature gradient and cooling rate within and around the melt pool, plays a critical role in microstructure evolution and cracking behavior. Nevertheless, existing research has mostly focused on composition optimization [24,25], and a systematic understanding of how the mesoscale thermal field distribution influences microstructure and cracking has not yet been fully clarified. In particular, it is unclear which region of the thermal field is most critical for promoting equiaxed grain formation and inhibiting crack propagation.

In situ thermal monitoring during LPBF, such as infrared thermography or pyrometry, can provide real-time temperature data [26,27]. However, these methods are often limited to surface measurements, suffer from low spatial resolution, and cannot capture the full temperature field inside the melt pool, especially its mesoscale distribution. Moreover, systematic experimental optimization involving numerous process parameters is time consuming and costly. Therefore, a well-calibrated finite element method (FEM) offers an efficient alternative to predict the thermal field and its evolution during LPBF [28,29].

In the present study, a finite element model is established to simulate the melt pool temperature fields of the original AA7075 and a 7075-Er-Zr modified alloy during LPBF. Key thermal parameters such as temperature gradient and cooling rate are extracted, and combined with microstructural characterization to systematically analyze the influence of the thermal field on solidification microstructures and crack formation in the two alloys. The results reveal a non-uniform distribution of temperature gradient within the melt pool, characterized by a peak-ring area. Interestingly, the annular region outside this peak ring contributes more significantly to crack suppression than the inner region. The role of Er and Zr addition in crack suppression under the influence of the thermal field is discussed, aiming to provide a theoretical basis for process optimization of LPBF-fabricated high-strength aluminum alloys guided by thermal field control.

2. Experiments and Modeling

2.1. Preparation of LPBF Samples

2.1.1. Materials

The AA7075 and 7075-Er-Zr alloy powders used in this study were provided by Beijing Crigoo Materials Technology Co., Ltd. (Beijing, China). The 7075-Er-Zr alloy was customized by adding approximately 0.55 wt.% Er and 0.35 wt.% Zr to the base AA7075 composition. The chemical compositions of both powders, listed in

Table 1. Chemical compositions of AA7075 and 7075-Er-Zr alloy powder (wt.%).

Elements	Si	Fe	Cu	Mn	Mg	Zn	Cr	Ti	O	Er	Zr	Al
wt.%	0.06	0.06	1.57	<0.02	2.53	5.9	0.21	0.005	0.046	/	/	Bal.
wt.%	0.09	0.85	1.74	0.02	2.39	4.2	0.31	0.086	0.079	0.65	0.29	Bal.

, were measured by inductively coupled plasma optical emission spectroscopy using a Thermo Scientific iCAP PRO X Duo instrument (Thermo Fisher Scientific, Waltham, MA, USA). As shown in the scanning electron microscopy (SEM) images in Figure 1a,b, both powders consist predominantly of spherical or near-spherical particles. Particle size distribution analysis indicates that the D₁₀, D₅₀, and D₉₀ values of the AA7075 powder are 15.21 ± 0.18 μm, 27.95 ± 0.02 μm, and 43.05 ± 0.28 μm, respectively, as shown in Figure 1c. For the 7075-Er-Zr powder, the corresponding values are 24.19 ± 0.08 μm, 43.76 ± 0.17 μm, and 70.09 ± 0.23 μm, as shown in Figure 1d. The laser transparency and reflectivity were measured using a PerkinElmer Lambda 750 UV/Vis/NIR spectrophotometer (PerkinElmer, Shelton, CT, USA). The laser absorptivity at 1064 nm was calculated to be 0.46 for the AA7075 powder and 0.49 for the 7075-Er-Zr powder.

2.1.2. LPBF Fabrication

Single-track configuration was adopted to reduce the influence of track overlap and layer-by-layer thermal accumulation, thereby allowing a clearer analysis of the relationship among processing parameters, melt pool thermal field, solidification microstructure, and crack formation. Powders on an AA7075 substrate were scanned using an SLM 125 HL system (SLM Solutions Group AG, Lübeck, Germany), equipped with a 1064 nm fiber laser delivering a maximum power of 400 W and a beam diameter of 80 μm. To minimize oxidation and contamination from interstitial elements, all samples were produced under a high-purity argon atmosphere with the oxygen content maintained below 0.1%. The detailed processing parameters are listed in

Table 2. LPBF processing parameters.

Process Parameters	Setting
Laser power (W)	200, 240, 280, 320, 360
Scanning speed (mm/s)	600, 800, 1000, 1200, 1400
Layer thickness (μm)	30
Laser spot diameter (μm)	80

. Five groups of single tracks for each parameter were printed.

2.1.3. Microstructure Characterization

After printing, the specimens were vertically sectioned by electrical discharge machining. The cross-sections were sequentially ground with SiC papers up to 2000 grit, followed by fine polishing with diamond suspension to obtain a smooth surface. For microstructural characterization, the polished surfaces were etched using Keller’s reagent (95 mL H₂O + 2.5 mL HNO₃ + 1.5 mL HCl + 1 mL HF) for 15 s, followed by rinsing with distilled water and ethanol. In contrast, samples for electron backscatter diffraction (EBSD) analysis were subjected to electropolishing in a solution of perchloric acid and ethanol (volume ratio 1:9) at 20 V, with a current maintained between 0.2 and 0.3 A. The electrolyte temperature was kept between −20 °C and −30 °C for a total polishing duration of 30 s. High-magnification microstructural imaging was performed using an Apreo S (Thermo Fisher Scientific, Waltham, MA, USA) field-emission scanning electron microscope, while EBSD diffraction patterns were collected through an HKL-EBSD system (Oxford Instruments, Abingdon, UK). The phase constitution of samples was examined by X-ray diffraction (XRD, Bruker D8 Advance, Bruker Corporation, Karlsruhe, Germany) using Cu Kα radiation at 40 kV and 40 mA. Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) were performed at 200 kV on a Tecnai G2 Spirit TWIN (Thermo Fisher Scientific, Waltham, MA, USA) using specimens prepared by ion milling with a Gatan 691 (Gatan, Inc., Pleasanton, CA, USA).

2.2. Numerical Model

2.2.1. Modeling and Meshing

A three-dimensional (3D) finite element model, comprising a solid substrate and a single powder layer, was developed in ANSYS APDL (ANSYS 2024 R2), as illustrated in Figure 2a. The substrate was defined as AA7075 with dimensions of 1.50 mm × 1.50 mm × 0.40 mm, while the powder layer measured 0.50 mm × 0.50 mm × 0.03 mm. To balance computational accuracy and efficiency, a transition meshing strategy with local refinement was employed within the melt pool region, with hexahedral elements utilized throughout the computational domain. The final mesh consisted of 83,286 nodes and 76,392 elements. Given the inherent complexity of the LPBF process, the following reasonable simplifications were adopted in the model.

(1)

The powder bed was modeled as a continuous and homogeneous medium. This homogenization approach is a common practice in LPBF thermal simulations for predicting mesoscale temperature fields [30,31].

(2)

The thermophysical properties of the materials were considered temperature-dependent, whereas the convective heat transfer coefficient and surface emissivity were assumed constant.

(3)

The laser absorptivity was experimentally measured for both powders and used as a constant in the FEM simulation.

(4)

Powder evaporation and the associated vapor recoil pressure were not considered. This simplification is further justified by the experimental observation that the LPBF process operated in the stable conduction regime without keyhole formation, where evaporation-induced effects are limited.

2.2.2. Governing Equation for Heat Transfer

The heat transfer in the LPBF process is governed by a 3D transient heat conduction equation, as shown in Equation (3) [32].

$${\displaystyle \frac{\partial }{\partial \mathrm{x}}}\left(k_x{\displaystyle \frac{\partial T}{\partial x}}\right)+{\displaystyle \frac{\partial }{\partial y}}\left(k_y{\displaystyle \frac{\partial T}{\partial y}}\right)+{\displaystyle \frac{\partial }{\partial z}}\left(k_z{\displaystyle \frac{\partial T}{\partial \mathrm{z}}}\right)+Q=\rho C{\displaystyle \frac{\mathrm{d}T}{\mathrm{d}t}}$$

(3)

where k_x, k_y, and k_z denote the thermal conductivities along the x, y, and z directions (W/m·K), T represents temperature (K), Q describes the volumetric heat input as a function of the fixed coordinates and time (J/m³), ρ is the density (kg/m³), C is the specific heat capacity (J/kg·K), and t is time (s).

The initial thermal condition at t = 0 s is defined as Equation (4).

$${T\left(x,y,z\right)}_{t=0}=T_0$$

(4)

where T₀ is set to the ambient temperature of 298.15 K. The heat transfer within the powder layer involves thermal conduction, convection, and radiation. The thermal boundary condition at the top surface is governed by a balance between the incident laser energy and convective and radiative losses to the surroundings. This balance is mathematically expressed by Equation (5) [33].

$$-k{\displaystyle \frac{\partial T}{\partial \mathrm{n}}}=Q\left(x,y,z,t\right)-h\left(T-T_0\right)-\mathsf{\epsilon }\mathsf{\sigma }\left(T^4-T_0^4\right)$$

(5)

where k denotes the thermal conductivity of alloy, n is the normal vector to the top surface of the powder bed, h is the natural convection coefficient, ε is the emissivity, and σ is the Stefan–Boltzmann constant.

2.2.3. Heat Source Model

The laser energy density in this model is described by a Gaussian profile in the x-y plane and an exponential decay along the z-axis, as illustrated in Figure 2b. For a laser beam scanning along the positive y-direction at a constant velocity v for time t, the transient thermal flux Q is defined as Equation (6) [34].

$$Q={\displaystyle \frac{2AP}{\mathsf{\pi }{R}^{2}H}}\exp \left(-{\displaystyle \frac{2\left({\left(\mathrm{x}-x_0\right)}^2+{\left(\mathrm{y}-y_0-v\mathrm{t}\right)}^2\right)}{R^2}}\right)\exp \left(-{\displaystyle \frac{\left|z\right|}{H}}\right)$$

(6)

where A denotes the laser absorptivity of the powder bed, P is the laser power, R is the laser radius, and H is the penetration depth of the laser into the powder bed.

2.2.4. Thermophysical Properties of Materials

In the simulations, the thermophysical properties are defined as a function of temperature, corresponding to the materials’ solid, liquid, or powder state. An element transitions permanently from a powder to a liquid state once its temperature surpasses the liquidus point, and subsequently solidifies upon cooling without reverting to the powder phase. The properties for the 7075-Er-Zr alloy were estimated by adjusting the well-established properties of the AA7075 baseline alloy to reflect its compositional modifications. The key thermophysical properties of materials, including density, thermal conductivity, specific heat, and experimentally measured laser absorptivity, are summarized in

Table 3. Thermophysical properties of AA7075 and 7075-Er-Zr.

Temperature (°C)	Density (kg/m3)		Thermal Conductivity (W/m·K)		Specific Heat (J/kg·K)		Laser Absorptivity
	AA7075/ 7075-Er-Zr		AA7075/ 7075-Er-Zr		AA7075/ 7075-Er-Zr		AA7075/ 7075-Er-Zr
25	2811	2826	14.0	14.7	860	856.7	0.46	0.49
100	2797	2812	16.1	16.9	901	894.7
200	2776	2792	17.5	18.4	944	937
300	2753	2769	18.9	19.8	984	977
400	2730	2746	19.9	20.9	1024	1017
500	2705	2721	30	31.8	1137	1130
600	2630	2681	50	52.2	5698	5733
700	2478	2496	181	188.5	1280	1273
800	2250	2260	220	229.1	1350	1343

[35,36].

3. Results and Discussion

3.1. Crack Characteristics

Figure 3 presents a one-to-one crack-distribution map extracted from the original optical micrographs of the top surfaces of single tracks for AA7075 and the 7075-Er-Zr alloy under different laser power and scanning speed combinations. The original optical micrographs are provided in the Supplementary Information (Figure S1), with enlarged views shown in Figure S2. In Figure 3, only melt-track boundaries and crack information are retained, which allows a clearer comparison of crack types and their distributions across processing conditions. The corresponding linear energy density E_L = P/v (where P is the laser power and v is the scanning speed) is annotated for each condition.

Two distinct crack morphologies are identified. The first type consists of long, wide longitudinal cracks propagating along the track centerline, marked by blue solid lines. Based on their location and orientation, these are characteristic of hot (solidification) cracks [37]. The second type comprises short, narrow, inclined lateral cracks extending between the melt pool boundary and the track center, indicated by red solid lines, most of which are hot cracks, while a few of which extend further into the substrate beyond the track edge as liquation cracks (indicated by green solid lines) [38]. Given the rarity of the liquation cracks and the difficulty in distinguishing them from hot cracks using only top-surface morphology, the present study does not separate the two crack types. Instead, all cracks are collectively considered for cracking susceptibility evaluation.

In Figure 3, dashed outlines in blue, orange, and red denote the predominant regions of different crack types: blue for inclined lateral cracks, red for centerline cracks, and orange for the coexistence of both. For AA7075, the red-outline region (centerline cracks) occupies a relatively large parameter area, especially at E_L above 450 J/m. In contrast, the 7075-Er-Zr alloy exhibits fewer continuous centerline cracks, and the blue-outline region (dominated by short inclined lateral cracks) becomes more pronounced. These results indicate that Er/Zr modification reduces crack propagation tendency and improves crack resistance of AA7075, although cracking is not completely eliminated.

The lengths of both lateral and centerline cracks were quantitatively measured from over ten micrographs acquired from different regions of the corresponding tracks using ImageJ v1.49t, as shown in Figure 4. A consistent trend is observed across all processing conditions: crack lengths in the 7075-Er-Zr alloy are significantly shorter than those in the AA7075 baseline. For example, at 200 W and scanning speeds 1000~1400 mm/s, inclined lateral cracks are completely suppressed in the modified alloy, while they persist with lengths of 30.3–52.7 μm per 100 μm in AA7075. With increasing linear energy density via higher power or lower speed, the crack length increases in both alloys, and longitudinal centerline cracking emerges. Notably, at 360 W and 800 mm/s, centerline cracks propagate continuously along the entire track in AA7075, but become discontinuous in the 7075-Er-Zr alloy. The data further reveal a power-dependent trend in both alloys: at lower powers (<280 W), inclined lateral crack lengths decrease with increasing speed, whereas at higher powers (>320 W), they peak at intermediate speeds (800~1000 mm/s), coinciding with the onset of significant centerline cracking. For longitudinal centerline cracks, their lengths generally diminish with increasing scanning speed. This quantitative evidence underscores the superior crack resistance imparted by Er/Zr microalloying.

The cross-sectional views in Figure 5 reveal the distribution of two crack types within the melt pools. The first type (blue arrows), corresponding to the centerline cracks in Figure 3, propagates along the melt pool centerline and is prevalent with E_L exceeding 300 J/m in AA7075 and 450 J/m in 7075-Er-Zr. The fewer occurrences of these centerline cracks in the cross-section than on the top surface suggest discontinuity along the scanning direction. The second type (red arrows), representing the cross-section of the inclined lateral cracks, appears along the pool sides. Their limited presence in these views indicates either restricted propagation or partial capture by individual cross-sections.

SEM analysis was performed to examine the characteristics of cracks and the microstructure, as shown in Figure 6. At low laser power (200 W, Figure 6a1–d1), no centerline cracks are evident. However, in the AA7075 samples (Figure 6a2,c2), small cracks at the melt pool bottom, hereafter referred to as microcracks, are observed (marked by white arrows). These microcracks are not discernible in the 7075-Er-Zr alloy (Figure 6b2,d2). The microstructures at the melt pool center and bottom show comparable grain sizes. When E_L increases to 600 J/m (360 W/600 mm/s, Figure 6e1–e4,f1–f4), centerline cracks become pronounced in both alloys, but are narrower in the modified alloy. Nevertheless, microcracks remain absent in the modified alloy with E_L of 600 J/m. In Figure 6e3,e4,f3,f4, the modified alloy exhibits finer grains compared to the coarse grains in the AA7075 alloy. Lateral cracks are rarely observed in these cross-sectional views, consistent with the sampling limitation discussed in Figure 5.

EDS composition analysis was performed near the cracks. As shown in Figure 7a, only a minor increase in silicon is detected adjacent to large cracks. For the centerline crack tail (Figure 7b), no clear silicon enrichment is observed; instead, a signal drop appears at the crack location. In the case of microcracks (Figure 7c), the crack is too narrow to resolve any compositional variation. This indicates that low-melting-point phases tend to segregate to the final solidification regions where wide cracks form, but such segregation is marginal and only detectable in relatively large cracks. Given the very low overall silicon content in the alloys, the observed silicon enrichment is subtle and not present in all crack types.

3.2. Microstructural Characterization

To further understand the effect of Er/Zr addition on grain structure and crystallographic texture, EBSD analysis was performed on samples fabricated at 360 W and 600 mm/s, as shown in Figure 8. The inverse pole figure maps reveal a clear difference in grain morphology between the two alloys. AA7075 is mainly characterized by columnar grains with relatively concentrated crystallographic orientations, whereas the 7075-Er-Zr alloy exhibits a finer and more equiaxed grain structure with more dispersed orientations (Figure 8a,d), as corroborated by the pole figures of the melt pool region. Grain boundary misorientation is a key microstructural indicator as well. Based on the misorientation angle θ, grain boundaries are typically classified into low-angle grain boundaries (LAGBs, 2° ≤ θ ≤ 15°) and high-angle grain boundaries (HAGBs, θ ≥ 15°) [39]. The grain boundary misorientation angle distributions (Figure 8b,e) show that the fraction of HAGBs increases from 59.97% in AA7075 to 65.80% in the 7075-Er-Zr alloy, indicating enhanced microstructural randomization. The grain size distributions further show that the average grain size decreases from 7.9 μm in AA7075 to 4.2 μm in the modified alloy (Figure 8c,f). These results indicate that Er/Zr modification promotes grain refinement and disrupts the preferential orientation of columnar grains.

XRD analysis reveals peaks consistent with Al₃Er and/or Al₃Zr phases, suggesting the possible formation of L1₂-structured precipitates, as shown in Figure 9. TEM analysis was further conducted, as shown in Figure 10. Figure 10a reveals a particle in the matrix, which are identified as Al₃Er and α-Al by the Fast Fourier Transform (FFT) images in the HRTEM (Figure 10b). The coherent orientation relationship (OR) between Al and Al₃Er was determined: (1$\stackrel{\mathrm{ˉ}}{1}\stackrel{\mathrm{ˉ}}{1}$)_α-Al//($\stackrel{\mathrm{ˉ}}{1}$11)_Al3Er, [011]_α-Al//[011]_Al3Er (Figure 10c). The measured interplanar angles between (200) and (1$\stackrel{\mathrm{ˉ}}{1}\stackrel{\mathrm{ˉ}}{1}$) in α-Al, and between (100) and ($\stackrel{\mathrm{ˉ}}{1}$11) in Al₃Er, are consistent with the theoretical values for FCC and L1₂ structures, respectively, further confirming the presence of the Al₃Er phase (Figure 10d). The interplanar distances calculated via inverse Fast Fourier Transform (IFFT) images indicated that the (1$\stackrel{\mathrm{ˉ}}{1}\stackrel{\mathrm{ˉ}}{1}$) interplanar spacing of α-Al matrix and L1₂-Al₃Er was 0.241 and 0.243 nm, respectively. A coherent interface is clearly visible in Figure 10d. Although the presence of Al₃(Er, Zr) was not accessed in this study, it was widely reported [40]. Based on the XRD and TEM analysis, together with the literature [41], Al₃Er, Al₃Zr, and Al₃(Er, Zr), are all considered responsible for the grain refinement in the modified alloys.

3.3. Temperature Field Analysis

Having addressed the refinement mechanism, it is noted that while microcracks are completely suppressed in the modified alloy, centerline cracks are not fully eliminated. This indicates that under the present processing conditions, microstructural refinement alone is insufficient to fully suppress centerline cracking, and the thermal field appears to play a significant role. Therefore, numerical simulations were performed to analyze the temperature fields of both alloys under different processing parameters.

To evaluate the reliability of the thermal simulation, the simulated melt pool width and depth were compared with the experimentally measured values obtained from cross-sectional views. It should be noted that the dimensions compared are the width W and depth D indicated in Figure 11a,b. The protruding region of the track is prone to artifacts from sectioning, polishing, or local damage during sample preparation. Therefore, the melt pool region within the substrate served as the reference. As plotted in Figure 11c,d, the simulated values are in good overall agreement with the experimental results within the selected processing parameter range. For the AA7075 alloy, the relative errors for melt pool width and depth range from 1% to 4% and from 3% to 11%, respectively. For the 7075-Er-Zr alloy, the corresponding relative errors are 1%–7% and 0%–14%, respectively. These error levels are comparable to those reported in previous LPBF thermal simulation studies, where melt pool depth errors typically range from 5% to 15% [42,43]. It is also noted that within the experimental parameter range investigated, no keyhole-mode melt pool morphology was observed. Therefore, despite the various simplifications made in the model, e.g., neglecting evaporation and vapor recoil pressure, the simulation results are considered sufficiently accurate.

Figure 12 presents the simulated cross-sectional temperature distributions, with melt pool boundaries marked by black dashed lines. The results show that melt pool dimensions (width and depth) increase with decreasing scanning speed or rising laser power. For AA7075 (Figure 12a1–e1,a2–e2), for example, the dimensions increase from 121 μm × 56 μm (200 W, 1400 mm/s) to 162 μm × 70 μm (200 W, 600 mm/s), and further to 229 μm × 104 μm (360 W, 600 mm/s). Under identical processing conditions, the 7075-Er-Zr alloy exhibits slightly larger melt pools, with minimum dimensions of 124 μm × 58 μm and maximum of 231 μm × 106 μm, attributable to its higher laser absorptivity.

Figure 13 reveals that the 7075-Er-Zr alloy exhibits higher melt pool depth to width ratios than AA7075 under identical processing conditions, with the ratio increasing with laser power and scanning speed. According to the literature [37,44], a higher depth-to-width ratio is often associated with increased solidification stress and higher cracking susceptibility. Nevertheless, the 7075-Er-Zr alloy displays superior crack resistance despite its less favorable melt pool geometry. This indicates that the beneficial microstructural optimization induced by Er/Zr addition effectively counteracts the negative influence of the less favorable melt pool geometry, with the microstructural effect dominating the cracking behavior.

The transient temperature profiles at the designated monitoring point on the top surface, crucial for solidification behavior, are shown in Figure 14a,b. The peak of each curve corresponds to the instant of laser passage, while the duration above the melting point defines the liquid lifetime. At a fixed scanning speed of 600 mm/s, increasing the laser power from 200 W to 360 W raises the peak temperature from 1555 °C to 2493 °C for AA7075 and from 1596 °C to 2573 °C for the 7075-Er-Zr alloy. Similarly, at a fixed laser power of 200 W, reducing the scanning speed from 1400 mm/s to 600 mm/s increases the peak temperature from 1482 °C to 1638 °C for AA7075 and from 1520 °C to 1681 °C for the 7075-Er-Zr alloy. These trends confirm that increasing energy input elevates peak temperatures and extends liquid lifetimes, leading to larger melt pools. Furthermore, the steeper slopes of the cooling curves indicate that higher laser power and scanning speed collectively result in increased cooling rates.

An analysis of the calculated cooling rates and temperature gradients at the monitoring point is presented in Figure 14c,d. As shown in Figure 14c, increasing the laser power leads to a moderate rise in cooling rates and temperature gradients for both alloys. In contrast, increasing the scanning speed resulted in a much more pronounced increase (Figure 14d). For instance, the cooling rate of AA7075 rises from 2.76 × 10⁶ °C/s to 1.21 × 10⁷ °C/s as the speed increases from 600 mm/s to 1400 mm/s, a change far exceeding that induced by varying the laser power, from 2.76 × 10⁶ °C/s to 5.22 × 10⁶ °C/s. This demonstrates that scanning speed is the dominant factor governing the thermal cycling rate and solidification cooling conditions, exerting a more significant influence on thermal gradients and cooling rates than laser power.

The spatial evolution of solidification microstructure is governed by the distributions of cooling rate and temperature gradient within the melt pool. Analysis of additional points (labeled in Figure 15a) reveals that both parameters generally increase with scanning speed at all locations (Figure 15b–e). Radially from the center to the boundary, cooling rates gradually decrease. The AA7075 shows significant spatial variation in cooling rate, whereas the 7075-Er-Zr alloy exhibits uniformly higher cooling rates with markedly reduced positional variation, indicating enhanced heat-transfer efficiency. The temperature gradient displays a non-uniform distribution, peaking at an intermediate radial position (Point 4) rather than at the center or edge. This pattern, attributable to the Gaussian energy profile of the laser beam and the fast heat transfer to the surroundings, creates a region of steepest thermal gradient. Remapping the temperature distribution (Figure 16) confirms this, showing the closest spacing of iso-temperature lines around Point 4. Furthermore, increasing the scanning speed expands the region between this peak gradient location and the melt pool boundary, a factor that critically influences the resultant microstructure distribution.

To predict the solidification morphology from the thermal field, the parameter G/R was calculated under the condition of 360 W and 600 mm/s using the simulated thermal fields. Values were extracted along the depth direction (Points 1–5) and transverse direction (Points 1, 6–9), as listed in

Table 4. Calculated G/R values at monitoring points under 360 W and 600 mm/s.

Direction	Point	G/R Values for AA7075 (×106 °C·s·m−2)	G/R Values for 7075-Er-Zr (×106 °C·s·m−2)
Depth direction	1	8.38	8.25
	2	32.56	30.26
	3	102.35	101.26
	4	282.11	297.17
	5	158.15	170.35
Transverse direction	1	8.38	8.25
	6	53.99	55.20
	7	138.48	141.20
	8	181.14	185.49
	9	166.47	170.42

.

As listed in Table 4, the G/R values are lowest at the melt pool center and increase toward the annular region. Along the depth direction, the maximum occurs at Point 4, 2.82 × 10⁸ °C·s·m⁻² for AA7075 and 2.97 × 10⁸ °C·s·m⁻² for the 7075-Er-Zr alloy. Transversely, G/R gradually increases from the center outward, remaining high near Points 8 and 9, with values of 1.81 × 10⁸ °C·s·m⁻² and 1.66 × 10⁸ °C·s·m⁻² for AA7075, and 1.85 × 10⁸ °C·s·m⁻² and 1.70 × 10⁸ °C·s·m⁻² for 7075-Er-Zr. These results suggest a stronger tendency for directional solidification in the region outside the melt pool center toward the annular periphery.

3.4. Influence of Temperature Field Distribution on Microstructure

To experimentally verify the influence of such thermal field distribution on microstructural evolution, EBSD analysis was performed on different regions of the melt pool. The melt pool was divided into four sub-regions, labeled a1-a4 for AA7075 and b1-b4 for the 7075-Er-Zr alloy (Figure 8). Pole figures extracted from these sub-regions are presented in Figure 17. Based on the local pole figures, the innermost regions (Figure 17a1,b1) exhibit relatively dispersed crystallographic orientations, whereas the outer regions (Figure 17a2–a4,b2–b4) show a more evident preferred orientation. For AA7075, this local difference is less pronounced because columnar epitaxial growth dominates across the melt pool, leading to relatively concentrated orientations in most regions. In the 7075-Er-Zr alloy, however, the orientation difference between the central and outer regions becomes more evident, underscoring the role of spatial variations in the temperature gradient. This indicates that, although Er/Zr modification promotes grain refinement and weakens the overall texture of columnar grains, local thermal conditions have a pronounced effect on grain orientation in different regions of the melt pool.

Figure 18 compares the HAGB fractions of regions Figure 18a1–a4 (AA7075) and Figure 18b1–b4 (7075-Er-Zr). For AA7075, the HAGB fraction decreases from 84.5% (Figure 18a1, center) to 78.2%, 74.1%, and 74.2% (Figure 18a2–a4, outer regions), indicating stronger orientation concentration toward the melt pool boundary. In contrast, the 7075-Er-Zr alloy maintains high HAGB fractions across all regions: 92.0% (Figure 18b1), 86.3% (Figure 18b2), 86.7% (Figure 18b3), and 88.9% (Figure 18b4). The outer regions (Figure 18b2–b4) have slightly lower values than the center but remain higher than their AA7075 counterparts. This suggests that Er/Zr modification weakens local orientation concentration in the outer annular region, and that spatial variations in the thermal field influence grain orientation distribution.

In welding and laser-directed energy deposition (L-DED), where high-power lasers (e.g., 10 kW) with millimeter-scale beams are used, the spatial variation in grain morphology across the melt pool is well established [45,46]. The CET model has been employed to describe this phenomenon [10,11]. It was also explained by two competing solidification mechanisms: heterogeneous nucleation on partially melted powder particles (promoting equiaxed grains) and epitaxial growth from the melt pool bottom (leading to columnar grains).

Such a spatially resolved microstructure distribution has rarely been reported in unmodified alloys fabricated by LPBF [10]. Consequently, the influence of the annular region on cracking has received little attention. In effective modified alloys such as Sc-containing high-strength aluminum alloys, a bimodal grain structure is typically observed. Primary Al₃Sc precipitates form near the melt pool boundary and act as nucleation sites for equiaxed grains, whereas the steep thermal gradient in the upper central region promotes columnar grain growth because nucleation sites are depleted there [47]. In such a bimodal structure, the equiaxed grain layer at the melt pool boundary can interrupt the continuous growth of columnar grains, thereby suppressing crack propagation even if columnar grains remain in the central region [48]. This explains why certain Sc-modified alloys exhibit improved crack resistance despite not being fully equiaxed, although a fully equiaxed structure is more desirable [49].

However, in the present 7075-Er-Zr alloy, even though a fully equiaxed grain structure is achieved throughout the entire melt pool, centerline cracks are still observed under high energy density conditions (e.g., 360 W, 600 mm/s). This seemingly contradictory result, having equiaxed grains yet showing persistent cracking, cannot be explained by the global equiaxed fraction alone, a metric commonly relied upon in studies [50,51]. Instead, the results reveal that the ring-shaped peak of the temperature gradient (Figure 16) creates an annular region where solidification is most directional. Within this region, despite an equiaxed morphology, the HAGB fraction is lower and the pole figures show a stronger texture than at the melt pool center. Such residual crystallographic orientation provides potential crack paths, making the annular region the critical weak link.

The identification of this annular region shifts the focus from bulk microstructure control to spatially targeted refinement. Under high energy density (e.g., E_L = 600 J/m), the severe thermal gradient in this region partially overrides the nucleation effect of Er/Zr, leading to coarser or more oriented grains that facilitate cracking. This finding suggests that further optimization should not merely aim for overall grain refinement, but rather specifically enhance nucleation within the annular region, for instance, by increasing the density of nucleation particles or by tailoring the energy input to reduce the temperature gradient peak (e.g., decreasing laser power or increasing scanning speed). More broadly, the concept of a thermally dictated critical annular region may be applicable to other crack-susceptible alloys in LPBF, offering a new perspective for process parameter optimization.

3.5. Crack Suppression Mechanism

In summary, this study elucidates the intrinsic relationship between the temperature gradient, solidification microstructure, and crack susceptibility in aluminum alloys processed by LPBF. The melt pool exhibits a characteristic temperature gradient distribution: high at the periphery, peaking in an intermediate annular region, and relatively lower at the center, as illustrated in Figure 19. Correspondingly, the cooling rate is higher in the central region than at the periphery. AA7075 develops coarse columnar grains in the peripheral region, which transition to relatively thinner columnar grains near the center. These aligned grains provide continuous paths for crack propagation, leading to severe centerline cracking. At the melt pool bottom, the coarse columnar grains grow in a less organized manner, leaving narrow intergranular gaps that appear as microcracks.

The introduction of Er/Zr modifiers fundamentally alters this microstructural evolution, as shown in Figure 19b. The resultant Al₃Er, Al₃Zr, and Al₃(Er, Zr) particles, which have been widely reported as effective nucleants in recent studies [40,52], significantly enhance the heterogeneous nucleation and lead to a refined equiaxed grain structure. Although equiaxed grains form throughout the melt pool, a steeper thermal gradient and a lower cooling rate near the boundary facilitate a higher degree of crystallographic orientation with larger grain size, whereas more random fine-grain orientations are observed in the central region. This refined and randomized microstructure disrupts the continuous grain boundary network, resulting in tortuous crack propagation paths. However, the microcracks were suppressed because the equiaxed grains and the randomized grain boundaries prevent the formation of continuous intergranular gaps.

4. Conclusions

This study integrates experimental characterization with thermal field simulation to elucidate the coupled effects of processing parameters and microstructure evolution on the cracking behavior of AA7075 and 7075-Er-Zr alloys during single-track scanning of LPBF. The principal findings are as follows:

(1)

Three characteristic crack modes are identified, showing a strong correlation with energy density. For AA7075, with E_L < 450 J/m, cracks propagated laterally. As energy density increases to above this value, deep centerline penetration becomes dominant. Microcracks are always observed at the bottom of the melt pool.

(2)

The 7075-Er-Zr alloy exhibits significantly improved cracking resistance. At E_L = 600 J/m, longitudinal centerline cracks still penetrate along the track. Crack-free fabrication is achieved at 200 W with scanning speeds above 1000 mm/s, while only short and discontinuous cracks form otherwise. Microcracks in 7075-Er-Zr are eliminated. The hot crack resistance enhancement is attributed to grain refinement and a transition to an equiaxed grain structure induced by microalloying, which randomizes grain orientation and disrupts long-range crack propagation.

(3)

A critical annular region dictated by the peak temperature gradient is identified in both alloys. In AA7075, this region contains aligned columnar grains that facilitate both microcracks and centerline cracks. In the 7075-Er-Zr alloy, microcracks are completely suppressed due to grain refinement and orientation randomization in this region. However, the annular region still retains a certain degree of crystallographic orientation, which grain refinement alone cannot fully eliminate. This residual texture is responsible for the continued occurrence of centerline cracks under high power densities.