Effect of Aging at Different Temperatures on Microstructure Evolution of 347H Heat-Resistant Steel-Welded Joints

Abstract

This study used 347H heat-resistant steel as the base material and systematically investigated the microstructural evolution and second-phase precipitation in typical regions during welding and aging processes. The results showed that the weld metal consisted of austenitic dendrites and inter-dendritic ferrite in a lath-like form. In the welded samples, the HAZ (Heat-Affected Zone) and BM (Base Material) regions were composed of equiaxed crystals. The microhardness of the HAZ was lower, mainly due to the coarser grain size and fewer second-phase particles. After aging at 700 °C, the hardness of all regions of the welded joint increased significantly due to the precipitation of M₂₃C₆ and MX phases. When the aging temperature increased to above 800 °C, the stability of the M₂₃C₆ phase decreased, and the diffusion rate of Nb in the matrix accelerated, promoting the preferential growth and stable presence of the MX phase. As the MX phase competes with the M₂₃C₆ phase for carbon during its formation, its generation suppresses the further precipitation of the M₂₃C₆ phase. Under 800 °C aging conditions, the γ/δ interface exhibited high interfacial energy, and the Nb content in the ferrite was higher, which facilitated the formation of the MX phase along this interface. As the aging temperature continued to rise, the hardness of the HAZ and BM regions initially increased and then decreased. After aging at 800 °C, the hardness decreased because the M₂₃C₆ phase no longer precipitated. After aging at 900 °C, the hardness of the HAZ and BM regions significantly increased, mainly due to the large precipitation of the MX phase. The hardness of the W (Weld Zone) and FZ (Fusion Zone) regions gradually decreased with the increase in aging temperature, mainly due to the reduction of inter-dendritic ferrite content, coarsening of second-phase particles, weakening of the pinning effect, and grain growth. In the 900 °C aging samples, the MX phase particle size from largest to smallest was as follows: W > HAZ > BM. The Nb-enriched ferrite provided the chemical driving force for the precipitation of the MX phase, while the δ/γ interface provided favorable conditions for its nucleation and growth; thus, the MX phase particles were the largest in the W region. The HAZ region, due to residual stress and smaller grain boundary area, had MX phase particle size second only to the W region.

1. Introduction

Against the backdrop of the rapid development of clean energy, Concentrated Solar Power (CSP) plants with thermal storage using molten salt are experiencing rapid growth [1]. Molten salt storage tanks and high-temperature medium transmission pipelines are key components in CSP systems and are typically made from 347H austenitic heat-resistant steel [2,3,4]. During the assembly and manufacturing process, these components usually require welding before being put into use, and their service life and operational safety largely depend on the reliability of the welded joints [5]. The performance of the welded joints is closely related to their microstructural evolution, mainly involving the stability of the second phase, the precipitation behavior of carbides, and the degree of grain coarsening [6]. The 347H heat-resistant steel exhibits excellent corrosion resistance and high-temperature mechanical properties in the solution-treated state, making it widely used in high-temperature service environments [7,8]. Among these, the MX phase is the primary strengthening phase in the steel, usually formed by Nb with C or N. This precipitate phase effectively improves the material’s creep resistance through a precipitation-strengthening mechanism and is an important factor in achieving its high-temperature strength [9,10,11].

However, there is a risk of age hardening during the service of heat-resistant steels. Research by Antunes, P.D. et al. [12] showed that aging treatment can improve the tensile strength, yield strength, and hardness of 317L welded joints but reduces elongation. Bai, J.M. et al. [13] also pointed out in their study of HR3C heat-resistant steel that long-term aging treatment at 700 °C significantly reduces the material’s elongation and impact toughness, primarily due to the precipitation of M₂₃C₆ carbides at grain boundaries. Chandra, K. et al. [8] found that after aging 347 heat-resistant steel at 650 °C, M₂₃C₆ carbides and σ phase precipitated at grain boundaries, significantly increasing the material’s sensitization.

Compared to the base metal, the weld metal exhibits a more pronounced tendency to embrittle after aging, and this trend intensifies with the increase in aging temperature [14,15,16]. Due to the relatively fast solidification rate during welding, microsegregation is likely to occur in the weld metal, accompanied by the formation of ferrite [17]. In austenitic stainless steel welds, an appropriate amount of δ-ferrite helps capture impurity elements, reduces the formation of low-melting phases, and effectively alleviates welding residual stresses [18]. However, studies by Bai, G. et al. [19] and Chandra, K. et al. [8] indicate that higher contents of δ-ferrite can increase the intergranular corrosion susceptibility of austenitic stainless steel, thereby posing a potential threat to the material’s service safety [8,19].

Research by Malhotra, D. et al. [20] shows that compared to 347 welds, 316L welds have poorer thermal stability of δ-ferrite, leading to easier carbide precipitation in the weld metal. At 550 °C, δ-ferrite gradually decomposes to form M₂₃C₆-type carbides; at temperatures above 650 °C, it further transforms into intermetallic compounds such as the σ phase, η phase, χ phase, and R phase. These precipitates predominantly distribute along the γ/δ interfaces and grain boundaries, significantly reducing the material’s toughness [21]. Luppo et al. [22] also confirmed that increasing the ferrite content in 347 stainless steel welds leads to a reduction in the material’s ductility. AB Rhouma et al. [23] conducted a long-term aging study on 316L welds in the 550–700 °C range, finding that δ-ferrite decomposed to form σ and χ phases, negatively impacting the mechanical properties of the weld joint. To mitigate the precipitation of M₂₃C₆ carbides and the σ phase, the ferrite content in the weld can be reduced, or Nb-containing filler wires can be used to improve the situation [23]. Moreover, Guan, K. et al. [24] compared the microstructures of 347 and 321 stainless steel welds after aging for 6000 h at 700 °C. The results showed that due to the presence of Nb, the number of M₂₃C₆ and σ phase precipitates in 347 stainless steel was significantly lower than that in 321 stainless steel.

Despite a large number of studies on the aging behavior of heat-resistant steels, systematic research on the organizational stability of 347H heat-resistant steel welded joints is still relatively scarce. Particularly at higher service temperatures, the influence mechanisms of the evolution of welded joint microstructures on the mechanical properties remain unclear. Therefore, this study focuses on two critical processes—welding and aging treatment—systematically analyzing the microstructure and hardness variations in typical regions of 347H heat-resistant steel-welded joints, providing theoretical support for the safety assessment of the material under high-temperature service conditions. Operating temperature is one of the core parameters determining the thermal-to-electric conversion efficiency of CSP power plants. Developing CSP plants capable of operating at ultra-high temperatures (typically in the 700–1000 °C range) is crucial for their broader application [25,26]. Since austenitic stainless steels containing Nb have already been commercially applied in environments up to 1123 K, this study selects the 700–900 °C temperature range for aging treatment [27].

2. Experiment Procedure

The 347H heat-resistant steel-welding samples were prepared from continuous cast billets, which underwent hot rolling and solution treatment at 1160 °C. The plate thickness was 20 mm, and its chemical composition is listed in

Table 1. Chemical composition wt.%.

Material	C	Si	Mn	P	S	Ni	Cr	Nb
Base metal	0.05	0.5	1.1	0.020	0.001	9.1	17.4	0.50
Weld wire	0.05	0.3	1.3	0.020	0.010	9.5	19.1	0.52

. The base material consists mainly of MX phases and equiaxed austenite grains, with a grain size of 6.9 ASTM and a low, uniformly distributed inclusion content. The welding was performed using the manual metal arc welding (SMAW) process for butt welding. Prior to welding, the edges of the base material plates were mechanically processed and pre-cleaned. Welding was strictly conducted following industrial standard procedures to ensure the quality of the welded joints. The welding filler material used was ER347 electrodes, which match the chemical composition of 347H (in compliance with AWS A5.9 ER347 standards) [28], with a diameter of 4 mm. Nine weld passes were performed during the welding process, with the welding current and voltage controlled at approximately 150 A and 30 V, respectively. The welding speed was 273 mm/min, the heat input was around 0.99 kJ/mm, and the interpass temperature was maintained below 150 °C. The welded samples were labeled as E.

Samples for aging treatment were cut perpendicular to the weld direction, with dimensions of 20 × 20 × 100 mm. The welded joints underwent aging treatment using a box-type resistance furnace at 700 °C, 800 °C, and 900 °C for 100 h. The resulting samples were designated as 700 °C, 800 °C, and 900 °C, respectively. After aging treatment, samples were extracted from the cross-section of the welded joints for microstructural observation. The specimens were ground and polished, followed by electrolytic etching with oxalic acid to reveal the microstructure of the weld metal. Additionally, a 60% nitric acid aqueous solution was used for electrolytic etching to delineate the grain boundaries. Furthermore, specimens were electrolytically etched in a 40% aqua regia solution for 5 s, with an average current density of approximately 1 A/cm². During this process, the matrix structure was etched away, while the precipitates within the specimen were retained.

The microstructure of the specimens was observed using a LEICA DM2700M (Leica Microsystems GmbH, Wetzlar, Germany) metallurgical microscope. A ZEISS Gemini 500 (Carl Zeiss AG, Oberkochen, Germany) field emission scanning electron microscope (SEM) was employed to examine the microstructure, and an ULTIM MAX 40 energy dispersive spectrometer (EDS, Oxford Instruments NanoAnalysis, High Wycombe, United Kingdom) was used to analyze the composition of secondary phases in the heat-resistant steel. The phase composition and grain size of the heat-resistant steel were determined using an Oxford Symmetry S2 electron backscatter diffraction (EBSD, Oxford Instruments NanoAnalysis, High Wycombe, United Kingdom) system. For EBSD analysis, the specimens were mechanically polished and then electrolytically polished in a solution containing 10% perchloric acid and 90% anhydrous ethanol. The EBSD scanning step size was set to 1.2 μm. Thermodynamic calculations were performed using the phase diagram and Scheil–Gulliver modules in the Thermo-Calc software (TCFE10 database, 2022a, Thermo-Calc Software AB, Solna, Sweden). According to the Chinese national standard “Metal Materials Vickers Hardness Test—Part 1: Test Methods” (GB/T 4340.1-2009) [29], the hardness test was performed with a test force of 9.8 N, a dwell time of 15 s, and an indentation spacing of 0.4 mm. Hardness measurements were conducted along three parallel lines for each weld joint sample, and the average value was calculated along with the corresponding error range.

3. Result and Discussions

3.1. Microstructure Analysis of Welded Joints

Low-magnification etching was performed on the cross-section of the welded joint before aging treatment, and the results are shown in Figure 1. It can be observed that the experimental steel has good welding quality, with no solidification defects, and the interface between the weld metal and the base material is clearly visible. Typically, the welded joint of austenitic heat-resistant steel can be divided into three regions: the weld metal, the heat-affected zone, and the base material.

The weld seam of the as-welded samples was observed under a metallurgical microscope (OM), and the results are shown in Figure 2. The weld microstructure exhibits typical dendritic features. Due to differences in cooling rate and heat flow, the dendritic morphology varies across different regions. As shown in Figure 2b,c, the weld structure consists of a white austenitic matrix and black ferrite between the dendrites. The austenite present in the microstructure at room temperature originates both from the direct solidification of the liquid metal and from the solid-state phase transformation of ferrite to austenite. Under faster cooling rates, ferrite typically forms in a lath-like morphology. During the ferrite → austenite phase transformation, diffusion is limited, and as the diffusion distance decreases, the tightly arranged lath structure becomes the effective mode of transformation. Therefore, the residual ferrite exhibits a lath-like morphology that cuts across the original dendrite or colony crystal growth direction [30]. As shown in Figure 2d, primary dendrites in the fusion zone grow preferentially along the direction perpendicular to the weld boundary, and merge at the center of the weld seam. Figure 2e,f shows the presence of planar dendrites in the interpass region between the weld beads.

Figure 3 shows the typical OM morphology of the BM and HAZ regions of the as-welded sample, with their grain sizes statistically analyzed. The results are shown in Figure 2c,d. The BM region mainly consists of equiaxed grains, with an average grain size of approximately 29.5 μm. Numerous fine black secondary phases are distributed on the austenite matrix, and the quantity of secondary phases in the BM region is greater than that in the HAZ region. During the welding process, the grains in the HAZ region undergo coarsening, with an average grain size of 47.6 μm, which is 61% larger than the grain size in the BM region. This phenomenon occurs because the heat-affected zone undergoes the welding thermal cycle, and the low thermal conductivity of the heat-resistant steel during the welding thermal cycle results in a significant amount of residual heat being stored, leading to grain growth in the HAZ.

Figure 4 shows the SEM morphology of the weld metal before aging treatment, with numerous secondary phases observed at the ferrite and austenite phase interfaces. EDS analysis at points 1 and 2 in Figure 4a indicates that these secondary phases are Nb-rich MX phases. The elemental distribution in the ferrite and austenite phases was statistically analyzed (Figure 4b), and the results show a significant difference in the contents of Cr and Ni elements between the two phases. Ferrite contains more Cr and Nb elements, while austenite has a higher content of Ni. As shown in Figure 4d, SEM morphology and EDS analysis reveal that ferrite contains more Cr and is more prone to corrosion. Additionally, at the δ/γ interface, many smaller Nb-rich MX phases are visible.

Figure 5 presents the SEM morphologies of different regions of the welded joint before aging. The area shown in Figure 5a corresponds to the interface between the BM and the weld metal, where the lower left region represents the HAZ, and the upper right is the FZ. The fusion zone is formed during welding by complete melting and solidification of the weld metal and adjacent base metal. As it consists of a mixture of filler metal and a portion of the base material that solidifies upon cooling, the fusion zone typically possesses a different chemical composition from the weld metal [31].

In the BM region, fine second-phase particles are dispersed primarily within grains, with a small number located at grain boundaries (Figure 5b). The quantity of precipitates in the HAZ is lower than in the BM, mainly due to the partial dissolution of precipitates during the welding process, as well as the partial elimination of grain boundaries. Austenite grains with clean boundaries begin to grow, leading to grain coarsening. No M₂₃C₆ precipitates were observed in any region of the as-welded sample. This is attributed to the suppression effect of Nb on M₂₃C₆ precipitation. It is well known that Nb exhibits a stronger affinity for carbon than Cr. During the welding thermal cycle, Nb preferentially combines with C and N to form MX-type precipitates.

3.2. Effect of Aging Temperature on the Microstructure of Welded Joints

3.2.1. Weld

Due to the heterogeneous nature of heat-resistant steel-welded joints, the microstructures in different regions exhibit variations in mechanical properties. When studying the effects of aging temperature on the microstructure of welded joints, it is essential to distinguish between the effects on the W zone, HAZ, and BM regions. Furthermore, the response to aging differs depending on the location. The optical microscopy images of the weld metal after aging at different temperatures are shown in Figure 6. The results indicate that after aging at 700 °C, the weld structure still consists of austenite and ferrite between dendrites, with the ferrite morphology similar to that before aging. After aging at 800 °C, the content of plate-like ferrite decreases, and the network-like distribution of ferrite is alleviated. Instead, short-bar and globular ferrite phases appear, arranged in parallel in a certain direction. After aging at 900 °C, the ferrite content significantly reduces, and the structure mainly consists of a complete austenitic matrix with second-phase precipitates at the grain boundaries and within the grains.

Welding is a non-equilibrium phase transformation process involving rapid local heating and cooling. During the solidification of the δ→γ phase transition, the temperature decreases too quickly, which restricts atomic diffusion, preventing the complete transformation of δ-ferrite. As a result, the structure at room temperature is unstable. When the sample undergoes aging at high temperatures, short-range atomic diffusion occurs, and the incompletely transformed δ-ferrite begins to undergo the δ→γ phase transformation [32]. The SEM images of the weld metal after aging at different temperatures are shown in Figure 7. The results reveal that after aging at 700 °C and 800 °C, the primary and secondary dendrite arms are relatively clean, with a small amount of secondary phases precipitating between dendrites. After aging at 900 °C, a large number of particle-like second phases appear in the weld region, with two sizes of precipitates present. This is attributed to the higher concentration of Nb in the ferrite phase, as well as the fact that Nb diffuses more rapidly in ferrite than in austenite, thereby accelerating the precipitation kinetics [33].

Figure 8 shows the SEM morphology of the MX phases precipitated in the weld metal after aging at 800 °C at different locations. At the δ/γ interface, the average size of the MX phase is 124.1 nm, while the MX phase in the austenite has a smaller average size of only 28.5 nm. The MX phase is primarily composed of C and Nb elements. Carbon has a high solubility in austenite and diffuses rapidly, whereas the diffusion of Nb is slower. As a result, the nucleation and growth of the MX phase are controlled by the diffusion of Nb. The energy at the γ/δ interface is higher, and the ferrite contains more Nb, making it easier for the MX phase to form along the γ/δ interface. The growth of carbides is attributed to the diffusion of carbon atoms from the austenite into the δ-ferrite, as well as the diffusion of Nb atoms from the δ-ferrite toward the δ/γ interface [23].

The Scheil–Gulliver module of the Thermo-Calc software was used to calculate the variation in Nb content in austenite and ferrite as a function of temperature during the solidification of 347H heat-resistant steel, as shown in Figure 9. The results indicate that the Nb content in ferrite is higher than that in austenite, which is consistent with the previous statistical findings. As the solidification temperature decreases, the content of Nb in both phases gradually increases due to segregation. At the end of solidification, the precipitation of Nb-rich MX phases causes a reduction in Nb content in both phases.

The ferrite content in the weld metal at different aging temperatures was statistically analyzed, and the results are shown in Figure 10a. After aging at 700 °C, the area fraction of ferrite remains nearly unchanged compared to the welded state. As the aging temperature increases to 800 °C, the area fraction of ferrite begins to decrease. When the aging temperature reaches 900 °C, the ferrite area fraction drops to 0.3%. The variation in alloying elements in the welded metal affects the formation of precipitates, which may lead to different mechanical properties of the welded metal. The composition of ferrite in the weld metal in the as-welded and aged states was also analyzed, as shown in Figure 10b. After aging at 800 °C, the Cr content in ferrite increased by 21%, while the Ni content decreased by 41%, as Cr from the austenite diffused into the ferrite. The Nb content in ferrite decreased from 0.5 wt.% to 0 wt.%. As mentioned earlier, the MX phase consumes more Nb from ferrite during its nucleation and growth process. When the concentration exceeds 19%, it may lead to the formation of a ferrite network, thereby reducing the corrosion resistance of the welded joint [33].

3.2.2. Heat-Affected Zone

The heat-affected zone (HAZ) has undergone changes in its microstructure and properties due to the specific welding thermal cycle. Figure 11 shows the OM morphology of the interface region between the weld metal and the base material at different aging temperatures. In the figure, the upper-left corner represents the fusion zone (FZ), and the lower-right corner represents the HAZ. The interface region shows that the grains at the edge of the fusion zone grow toward the center of the weld, extending into the weld metal. Although the HAZ does not undergo melting, severe overheating causes the dissolution and re-formation of precipitates, ultimately leading to grain coarsening. The temperature range for overheating is between 1100 and 1450 °C, during which austenite grains begin to grow. Typically, the plasticity and toughness of the welding heat-affected zone are lower, especially the impact toughness, which is 20–30% lower than that of the base material [34]. Garcia, C et al. [35] investigated the welded joints of austenitic stainless steels and found that the degree of sensitization in different welding regions is closely related to local variations in composition and microstructure during the welding process. Their study revealed that the heat-affected zone (HAZ) is the most susceptible region to intergranular corrosion. As the aging temperature increases, the microstructure of columnar grains gradually becomes more evident in the FZ region, and ferrite dissolves to form MX phases. In addition, the grain size in the HAZ region increases gradually, but the grain coarsening is not significant, indicating that 347H heat-resistant steel has high thermal stability under high-temperature conditions. From 700 °C to 800 °C, there is little change in the precipitates in the HAZ. However, as the aging temperature rises to 900 °C, the microstructure of both the HAZ and FZ regions undergoes significant changes, with both regions precipitating a large amount of fine second phases.

3.2.3. Base Metal

Figure 12 shows the SEM morphology of the base material after aging at 700 °C. The results indicate that numerous second phases are distributed on the austenitic matrix. Figure 12b in the upper-right corner presents the corresponding backscattered electron (BSE) image. BSE imaging reveals the atomic number contrast of the measured region, with higher atomic numbers, corresponding to brighter areas in the image. The BSE morphology of the second phase shows a bright contrast, indicating that these phases are MX phases, with an average diameter of 124.9 nm. The nanoscale dispersed MX phases act as effective obstacles to dislocation motion, significantly strengthening the 347H heat-resistant steel by pinning and suppressing dislocation movement. In heat-resistant steels, the precipitation of the MX phase is crucial for improving high-temperature endurance performance [36]. Additionally, elongated voids are observed at the grain boundaries, which may be sites where the M₂₃C₆ phase was removed during the electrochemical etching process in a 60% nitric acid solution.

Using 40% aqua regia for electrochemical etching, the morphology of M₂₃C₆ phase precipitates at the grain boundaries of the 700 °C aged sample was observed, as shown in Figure 13. M₂₃C₆ phases become visible at the grain boundaries of the aged test steel, appearing as long strips and almost entirely distributed along the grain boundaries. This is because carbon atoms rapidly diffuse at the grain boundaries via dislocations and vacancies, causing the M₂₃C₆ particles to nucleate and grow primarily at these boundaries. From Figure 13b, it can be seen that the size of the M₂₃C₆ phase is significantly larger than that of the MX phase. In the BSE morphology in the upper-right corner of Figure 13b, three contrasts are present: the brightest particulate MX phases precipitated within the grains and at the grain boundaries, the darker M₂₃C₆ phase, and the γ phase. The brightness of the M₂₃C₆ phase in the BSE mode is lower than that of the MX phase, and it is closer to the brightness of the austenitic matrix.

From Figure 13c,d, it can be observed that two types of precipitates exist at the grain boundaries: one is the Nb-rich MX phase, and the other is the Cr-rich M₂₃C₆ precipitate. Fukunaga, T. et al. [37] conducted a study on the formation of intergranular M₂₃C₆ in 347 stainless steel. The results showed that at the NbC/matrix interface, the Nb content in the matrix nearly dropped to zero and then gradually increased. With the extension of aging time, the width of the low Nb content region increased. The regions with lower Nb content may serve as nucleation sites for Cr-rich M₂₃C₆, especially near NbC and along grain boundaries, where they grow in the austenitic matrix. The free carbon provided by the matrix promotes this process. As the Nb content decreases, the formation kinetics of M₂₃C₆ become more favorable. Figure 13e shows the typical morphology of M₂₃C₆, with point 1 indicating the Cr- and C-rich M₂₃C₆ phase and point 2 indicating austenite. M₂₃C₆ carbides are primary harmful precipitates that form during aging, with a complex face-centered cubic structure. Although M₂₃C₆ is a metastable phase, it easily nucleates and appears early in the precipitation process. When enough M₂₃C₆ precipitates, the matrix becomes depleted in chromium, leading to intergranular corrosion.

Figure 13d shows that the average size of M₂₃C₆ is 500.0 nm, which is four times the size of the MX phase. Studies have suggested that the fine, discontinuous M₂₃C₆ precipitates at the early stages of aging may improve the resistance to grain boundary sliding in 347H and contribute to enhancing its creep strength [38]. However, due to the faster coarsening rate of M₂₃C₆ compared to MX, the quantity and size of M₂₃C₆ will significantly increase over the same aging time [39]. This can be attributed to two factors: first, carbon atoms diffuse more rapidly near grain boundaries, which facilitates the formation of M₂₃C₆ precipitates. Additionally, the dislocations and vacancies distributed along the grain boundaries also promote the nucleation of M₂₃C₆. Second, M₂₃C₆ has a coherent relationship with the matrix, which lowers the nucleation energy. M₂₃C₆ carbides precipitate sequentially at the grain boundaries, non-coherent twin boundaries, and coherent twin boundaries, eventually precipitating inside the grains [40].

3.3. Thermodynamic Calculations

The high-temperature performance of 347H heat-resistant steel is closely related to the formation of various precipitates under high-temperature service conditions. The stable phase content of the experimental steel in the 500–1500 °C temperature range, calculated using Thermo-Calc software, is shown in Figure 14. The results indicate that the matrix structure is austenite, with the primary precipitates being σ, MX, Laves, M₂₃C₆, and Z(NbCrN) phases.

When austenitic steel containing Nb also contains sufficient nitrogen (N), it is possible to form the Z phase, which is a carbide-nitride with a tetragonal structure. During long-term aging or prolonged high-temperature operation, the MX-type precipitate can transform into the Z phase [41]. At 1370 °C, MX (FCC_A1#2) begins to form. As MX precipitates in the austenite, it can pin the austenite grain boundaries, thereby hindering grain boundary migration and contributing to grain refinement. When the temperature continues to decrease to 800 °C, precipitates such as M₂₃C₆ and the Z phase begin to form.

3.4. Quantitative Analysis of the Second Phase

3.4.1. Second Phase at Different Aging Temperatures

The SEM morphology of the base material after aging at 800 °C and 900 °C is shown in Figure 15. The results indicate that dispersed, particle-like MX phases are present in the austenitic matrix, while no M₂₃C₆ phases are observed. According to the previous thermodynamic calculation results, M₂₃C₆ will only precipitate when the temperature decreases to 796 °C, meaning that aging at temperatures higher than this will prevent the formation of M₂₃C₆. M₂₃C₆ is a metastable phase, and its thermodynamic stability is poor at high temperatures. As the aging temperature increases, the free energy difference of M₂₃C₆ decreases, tending to evolve towards a more thermodynamically stable state, forming a more stable MX phase. Additionally, at higher temperatures, the diffusion rate of Nb in the matrix increases, promoting the preferential growth and stabilization of the MX phase. The formation of the MX phase competitively consumes carbon, thus limiting the precipitation of M₂₃C₆ [8]. Glazoff, M.V. et al. [42] simulated the precipitation of phases in 347H at 750 °C, and the results showed that M₂₃C₆ completely dissolved after 120 h.

Stabilization treatments of the alloy (such as heating to 850–1000 °C and holding for 2 h) allow elements that form strong carbides, like Nb, V, and Ti, to preferentially combine with carbon, forming MC carbides. Higher aging temperatures accelerate the nucleation and growth of MX carbides, which reduces the carbon content in the austenitic matrix, thereby inhibiting the formation of M₂₃C₆ at the grain boundaries [37]. Xia, Z.X et al. [43] showed that under intermediate heat treatment conditions at 850 °C, the first precipitates formed in the steel are MX carbonitrides rather than M₂₃C₆ carbides, which results in a decrease in the carbon concentration in the matrix, thereby reducing both the volume fraction and average size of M₂₃C₆.

As shown in Figure 15c,f, with the aging temperature increasing from 700 °C to 800 °C, the average size of the MX phase increases from 124.9 nm to 134.6 nm, but the coarsening trend is not significant. When the aging temperature is further increased to 900 °C, the size of the MX phase increases to 160.4 nm. The statistical results (see

Table 2. Statistics of the number density of MX phase at different aging temperatures.

Samples	Number Density/μm2
700 °C	0.44
800 °C	0.59
900 °C	0.91

) show that aging temperature significantly affects the quantity of precipitates in 347H heat-resistant steel. As the aging temperature increases, the number density of MX phases increases. Higher isothermal temperatures enhance the diffusion rate of atoms, and the increase in temperature accelerates the coarsening rate of carbonitrides.

3.4.2. Second Phase in Different Regions of Welded Joints

Figure 16 shows the morphology of the MX phase in the weld zone of the sample aged at 900 °C. A large number of particle-like secondary phases have precipitated on the austenitic matrix. EDS analysis (Figure 16b) reveals that these secondary phases are MX phases rich in Nb. As shown in Figure 16d, the average size of the MX phase is 300.6 nm, which is an 87% increase compared to the MX phase in the BM. The growth process of the precipitates is mainly influenced by diffusion processes and interface energy [44]. Since the diffusion rate of Nb is relatively slow, the growth of the MX phase is typically controlled by the diffusion process of Nb. The concentration of Nb in the ferrite is higher, and its diffusion rate in ferrite is greater than in austenite, which provides the chemical driving force for the precipitation of the MX phase. The δ/γ interface has a high interface energy, and when the MX phase forms at this interface, the free energy is lower, thus providing favorable sites for the nucleation and growth of the MX phase. Therefore, the size of the MX phase is the largest in the weld zone [23]. Research by Malhotra, D et al. [20] shows that the interdendritic regions in the weld metal are the most favorable sites for carbide formation. Aging at higher temperatures accelerates the decomposition process of δ-ferrite.

Figure 17 shows the SEM morphology of the interface between the weld metal and the base material in the sample aged at 900 °C. A large number of MX phases have precipitated in the HAZ. As shown in Figure 17b,c, the sizes of the MX phases precipitated along the grain boundaries in the HAZ are larger than those within the grains. Compared to the base material, the HAZ contains more MX phases at the grain boundaries because the grain boundaries have higher interface energy and provide a faster diffusion path for alloying elements. During welding, the HAZ is heated to temperatures near the alloy’s solidus, causing many of the precipitates in the base material to dissolve. This results in an oversaturation of the austenitic matrix during cooling, leading to the formation of various precipitates, which typically form along the grain boundaries. The residual stresses in the welding heat-affected zone cause the MX phases to nucleate along dislocations, making the HAZ region more prone to growth.

As shown in Figure 17d, the average size of the MX phase in the HAZ is 265.7 nm, which is slightly smaller than the size of the MX phase in the weld zone. The size order of the MX phases across the different regions is as follows: weld zone > HAZ > BM. The grain size in the BM region is finer, and the grain boundary area is larger, providing more nucleation sites, which leads to the smallest MX phase size. Studies show that rapid cooling at welding temperatures leads to residual stresses in the MX phases in the heat-affected zone, triggering precipitation hardening. During high-temperature creep service, the coarser MX phases at the HAZ grain boundaries tend to form creep voids. Under relatively low-temperature, high-stress conditions, failure typically occurs in the ferrite of the base material or the weld metal; however, under low-stress conditions within a specific temperature range, failure generally occurs near the HAZ [45].

3.5. EBSD Analysis

To further investigate the microstructure of different regions in the welded joint, this study employed EBSD (Electron Backscatter Diffraction) to comprehensively characterize the crystallographic information of the 700 °C aged sample. Figure 18a presents an Inverse Pole Figure (IPF) showing the grain orientation distribution. The upper left corner represents the fusion zone (FZ), which exhibits a columnar grain structure, while the lower right corner represents the heat-affected zone (HAZ), which exhibits an equiaxed grain structure. Understanding the nature of different boundaries or interfaces in the heat-resistant steel weld metal is crucial because many defects that occur in the weld during manufacturing and service are closely related to these boundaries. Figure 18b shows the distribution of large-angle and small-angle grain boundaries at the interface between the base metal (BM) and the weld. In the figure, grain boundaries with misorientation angles less than 2° are represented by black lines, while grain boundaries with misorientation angles between 2° and 15° are defined as low-angle grain boundaries (LAGB) and are represented by green lines. Coincidence site lattice (CSL Σ3) boundaries are shown in the figure using red lines.

In the entire interface region, large-angle grain boundaries dominate, accounting for as much as 94.4%. In the fusion zone (upper left corner of the image), the solidification grain boundaries are formed by the intersection of sub-grain bundles or sub-grain groups. These solidification grain boundaries result from the competitive growth of grains along the pool boundary during the solidification of the weld pool. Since each sub-grain bundle has different growth directions and lattice orientations, the intersection of these bundles forms large-angle grain boundaries, which become the primary grain boundary type in the weld. In austenitic stainless steels, weld solidification cracks often form along the sub-grain boundaries (SGB) [17,46].

In the lower right corner of the image, the heat-affected zone (HAZ) is displayed with small grains. The HAZ region shows a higher proportion of coincident site lattice (Σ3) boundaries, which is closely related to the hot-rolling and solution-annealing treatment of the base metal. Twin boundaries play an important role in austenitic steels, not only helping to improve ductility and reduce the likelihood of brittle fracture but also enhancing the intergranular corrosion resistance of heat-resistant steels [47].

To analyze the deformation degree of the welded joint, the local average misorientation (KAM) method is typically employed. KAM calculates the orientation of a given pixel point within a grain and compares it with the orientation of the six surrounding neighboring points to assess the local plastic strain or deformation within the grain [48]. Figure 18c shows the KAM distribution map, where stress is mainly concentrated at the interface of the fusion zone and HAZ, although the overall stress level is relatively low, with an average KAM value of 0.26°.

3.6. Microhardness Distribution

Microhardness is generally considered a reflection of the microstructure. Figure 19a shows the hardness distribution curve of the welded joint, while Figure 19b presents the hardness distribution across the cross-section of the 347H welded joint, from the weld metal to the base material. The statistical results of the average microhardness values for each part of the welded joint in different states are shown in

Table 3. Hardness values (HVs) of different regions of the welded joint.

Samples	W	FZ	HAZ	BM
E	188.8	199.4	191.1	204.5
700 °C	220.4	224.2	214.8	224.3
800 °C	205.7	212.4	204.2	210.1
900 °C	183.0	184.5	234.9	233.4

. For the as-welded samples, the order of hardness values is as follows: BM > FZ > HAZ > W. The microstructure of the BM region is finer than that of the HAZ region, so, according to the Hall–Petch relationship, the hardness in the BM region is higher. The microhardness in the HAZ is lower, mainly due to its relatively coarse grains and fewer second-phase particles. The HAZ near the fusion boundary is a coarse-grained heat-affected zone (CGHAZ) and has lower hardness, while the HAZ near the base material is a fine-grained heat-affected zone (FGHAZ) and has higher hardness. This trend persists after aging treatment. The region near the fusion zone forms coarse grains due to the slower cooling rate, while the area near the base material forms fine grains due to the larger thermal gradient and faster cooling rate. The fusion zone exhibits higher hardness, consistent with previous research results [49]. This is primarily attributed to some unmelted grains in the FZ, which act as nucleation sites for new phase precipitates during the solidification of the weld metal. Additionally, due to the higher ferrite content in the FZ, the precipitation of carbide particles also hinders the propagation of dislocations across grain boundaries, thus increasing the hardness of the material.

The hardness distribution exhibits different trends with varying aging temperatures. After aging at 700 °C, the hardness values in all four regions increase significantly. The hardness in the W region rises from 188.8 HV in the as-welded state to 220.4 HV, while the hardness in the BM region increases from 204.5 HV to 224.3 HV. This change is attributed to the precipitation of M₂₃C₆ and MX phases. The second-phase particles hinder the dislocation slip under the indentation force, thereby increasing the microhardness. As the aging temperature increases, the hardness in the W and FZ regions gradually decreases, primarily due to the reduced ferrite content between the dendrites, grain growth, and the coarsening of second-phase particles. In the sample aged at 900 °C, the W region achieves the lowest hardness value. The ferrite in the W region dissolves, the precipitates aggregate and grow, and the grain size increases, leading to a reduction in hardness. Research by Omiogbemi, I.M.B. et al. [50] indicates that the ferrite content between dendrites plays a crucial role in hardness in the AISI 2205 duplex stainless steel-welded joints. The δ-ferrite in the weld enhances hardness through second-phase strengthening, and the δ-ferrite precipitated along the grain boundaries helps promote the pinning effect, suppressing the growth of austenite grains and thereby improving the strength of the welded joint [51].

As the aging temperature increases, the hardness in the HAZ and BM regions first increases and then decreases. The hardness variation is directly related to the number of carbides precipitated in the microstructure; the more precipitates, the higher the hardness. After aging at 800 °C, M₂₃C₆ no longer precipitates, leading to a reduction in the overall number of precipitates and a corresponding decrease in hardness. However, after aging at 900 °C, the hardness in the HAZ and BM regions significantly increases, mainly due to the large amount of MX phase precipitation and the increase in its size. The second-phase particles more effectively hinder dislocations, thus increasing the overall microhardness.

4. Conclusions

• The weld structure of 347H consists of austenitic dendrites and interdendritic ferrites in the weld zone. The HAZ and BM in the as-welded sample are composed of equiaxed grains. The average grain size in the HAZ region is 61% larger than in the BM region. The microhardness ranking is as follows: BM > FZ > HAZ > W. The microhardness in the HAZ region is lower, primarily due to the relatively coarse grains and fewer second-phase particles.
• After aging at 700 °C, the hardness values in all regions of the weld joint increase significantly due to the precipitation of M₂₃C₆ and MX phases. The size of M₂₃C₆ in the base material is four times that of the MX phase. When the aging temperature exceeds 800 °C, the stability of the M₂₃C₆ phase decreases, and the diffusion rate of Nb in the matrix increases, allowing the MX phase to preferentially grow and remain stable. The formation of the MX phase competitively consumes carbon, thus limiting the precipitation of M₂₃C₆6.
• After aging at 800 °C, the size of the MX phase in the weld zone austenite is significantly smaller than that of the MX phase at the γ/δ interface. The nucleation and growth of the MX phase are controlled by the diffusion of Nb. Due to the higher energy of the γ/δ interface and the higher Nb content in ferrite, the MX phase readily forms along the γ/δ interface, consuming Nb from the ferrite.
• As the aging temperature increases, the hardness in the HAZ and BM regions first increases and then decreases. After aging at 800 °C, the hardness decreases because M₂₃C₆ no longer precipitates. After aging at 900 °C, the hardness in the HAZ and BM regions increases significantly due to the large amount of MX phase precipitation and its increased size. As the aging temperature rises, the hardness in the W and FZ regions gradually decreases, mainly due to the reduction in ferrite content between dendrites, coarsening of second-phase particles, weakening of the pinning effect, and grain growth.
• After aging at 900 °C, the size of the MX phase follows the order: W > HAZ > BM. Nb in ferrite provides the chemical driving force for the precipitation of the MX phase, and the γ/δ interface provides favorable sites for the nucleation and growth of the MX phase. Therefore, the MX phase size is the largest in the W region. Due to the residual stress and smaller grain boundary area in the HAZ region, its MX phase size is second only to that in the W region.