Abnormal grain growth under solution-aging treatments enables convergence mechanical properties of a Fe-Ni-Cr based austenitic alloy

In this study, systematic heat treatment routes including solution (980 °C to 1130 °C) and aging (675 °C to 775 °C) processes were used to tailor the microstructural evolution and mechanical performance of a Fe-Ni-Cr based austenitic alloy, strengthened by a γ′ phase (Ni3(Ti, Al)). Grain growth was observed with increasing solution treatment temperature from nearly 90 μm (980 °C) to nearly 200 μm (1130 °C). Grain refinement during solution heat processes was found to ascribe to the solute drag effect and the pinning effect of nickel-titanium-enriched segregates and precipitates. During aging, the precipitation behavior of γ′ is found to be almost independent of solution treatment temperatures. Interestingly, abnormal grain growth during aging was observed. The motion of grain boundaries was ascribed to the formation of nickel-titanium-enriched γ′ in the austenitic matrix, thereby contributing to the dissolution of the previously enriched segregates and precipitates from grain boundaries. The studied alloy shows a wide range of mechanical properties, from tensile strength of 1131 MPa with ductility at 36%, to tensile strength of 880 MPa with ductility at 50%. The current study demonstrates that grain refinement of the alloy at solution treatment may not benefit the final mechanical properties.


Introduction
High-pressure, high-temperature liquid rocket engine components, containing liquid and gaseous hydrogen propellants, provide a challenging operating environment for materials [1][2] [3]. At present, Fe-Ni-Cr based alloys JBK-75 [3] and A-286 [4] are commonly used in such applications, owing to a γ-phase matrix that is highly resistant to hydrogen embrittlement [5]. They are further strengthened by a nano-ordered coherent γʹ phase (Ni 3 (Ti, Al)) that is homogeneously distributed in γ-matrix [6]. However, JBK-75 and A-286 are susceptible to generating an hcp η phase (Ni 3 Ti) [7], which deteriorates mechanical capabilities [8] and hydrogen resistance [9]. The A-286 alloy also has low weldability due to hot cracking [10] and is not compatible with current additive manufacturing techniques. To correct the aforementioned problems, a novel Fe-Ni-Cr based austenitic alloy, NASA-HR (NHR), was developed [11]. Compared with A-286 and JBK-75, the improvements of this new alloy include: (1) the ability to systematically modify γ matrix compositions; (2) the increased γʹ phase volume fraction and added γ-matrix strengthening elements; and (3) the retarded precitipitation of the η phase on grain boundaries (GBs) [12]. To utilise knowledge from traditionally hydrogen-resistant alloys, tungsten and molybdenum were added to the new alloy to stabilise GBs and restrain η phase precipitation. Therefore, NHR combines high yield strength, corrosion/oxidation resistance, and resistance to hydrogen embrittlement [13].
For NHR and other γʹ phase precipitation-strengthened alloys, the strength contribution stems mainly from grain refinement and precipitation strengthening [14]. The Hall-Petch relationship describes the well-known increase in strength of metallic materials as grains decrease in size [15]. To obtain a smaller austenite grain, alloys Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
such as Inconel 718 [16] and A-286 [17] are often treated with a low-temperature solution heat treatment, while ensuring complete recrystallization. Precipitation strengthening can be achieved by producing a dispersion of particulates that act as obstacles to dislocation movement. Aging at a lower temperature facilitates the formation of a precipitation phase from the supersaturated austenite matrix. This is a fine balance; a significantly low aging temperature will fail to stimulate nucleation and growth of precipitates, which limits the effects of preciptation hardening on an alloy's strength. However, aging at higher temperatures may activate the formation of a detrimental phase at GBs. Thus, the solution-aging treatments of the NHR alloy should be carefully designed, and the influence of the parameters on final mechanical properties should be clarified.
Grain growth is generally observed as a temperature-dependent process in solid-solution heat treatments of austenitic alloys. This has been widely explained by thermally activated atomic jump processes [18,19]. The drag effect of solute atoms in the matrix and the pinning effect of precipitates also significantly affect grain growth [20]. Cahn [21] proposed the solute drag effect, describing hindered GB movement due to solute atoms segregating at GBs. Lan et al [19] found that molybdenum and niobium have large lattice mismatches with the Inconel 718 matrix, causing an accumulation of defects at GBs, further impeding their movement. Zener and Smith's [22] theory on the pinning effect expects that the precipitates may exerts pressure on GBs and therefore impedes grain growth. The effect of solute drag or pinning has been studied with relation to the motion of GBs at same thermal conditions [23]. However, for austenitic matrix, the effect of the new formed precipitates at a different thermal condition on the GBs stabled at the previous heat treatment has not been noticed. The newly formed precipitates could potentially break the existing chemical balance between the matrix and the GBs, thus enable the further motion of the GBs.
In this study, a systematic heat treatment, including both solution and aging processes, is used to tailor the microstructural evolution and mechanical performance of NHR. Interestingly, abnormal grain growth is observed during low-temperature aging treatments. Various characterization methods such as scanning electron microscopy (SEM), electron backscatter diffraction (EBSD), secondary ion mass spectroscopy (SIMS), and transmission electron microscopy (TEM) were used in this study in order to characterise grain growth, precipitates, and elemental distribution within the alloy. The study elucidates the mechanism of the interaction between grain growth and precipitation hardening, and finally suggests an optimal heat treatment route for NHR alloys. It is expected that the grain-refinement during solution process may not benefit the mechanical properties of the studied alloy.

Materials and experimental methods
The chemical composition (wt %) of the studied Fe-Ni-Cr based (NHR) alloy is listed in table 1. Ingots were melted in a vacuum induction furnace and then forged into bars with diameters of 80 mm. The tested heat treatment routes are displayed in figure 1. Since the η-phase has been previously examined to exist in the alloy at temperatures between 800°C and 930°C, the solution treatment temperatures were chosen above 980°C to eliminate the η-phase. Similarly, temperatures below 775°C were selected for the aging treatment in order to achieve sufficient generation of the strengthening phase (γ′). Heat treatments were carried out at solution treatment temperatures of 980°C, 1030°C, 1080°C, and 1130°C for 1.5 h, followed by oil quenching. Then the alloys were aged at 675°C, 700°C, 725°C, 750°C and 775°C for 8 h, followed by air cooling. The samples are hereafter marked with an 'S' for solution temperature, and an 'A' for aging temparature, such as S980, or S1130A750.
As shown in figure 2(a), the specimens were machined into tensile samples with gauge size of Φ3 mm×15 mm. Tensile tests were conducted using a GNT300 electronic tensile testing machine equipped with an extensometer with a constant cross head velocity (2.5 × 10 −4 mm s −1 ) at room temperature. Impact tests were performed with U-shaped notch specimens (2.5 × 10 × 55 mm 3 ) at room temperature ( figure 2(b)).
Multiscale characterization techniques were applied to investigate the microstructural evolution of NHR at different states. These include an Olympus GX53 optical microscope (OM), an FEI Quanta 650 FEG scanning electron microscope (SEM), electron backscatter diffraction (EBSD), an FEI Talos transmission electron microscope (TEM), and TOF-secondary ion mass spectroscopy (SIMS). Specimens for OM and SEM were mechanically polished and ground, and then chemically etched in a CuCl 2 mixture (CuCl 2 + HCl + H 2 O + C 2 H 5 OH). The fracture morphologies were directly observed with SEM. EBSD measurements were performed using an accelerating voltage of 20 kV, a step size of 1 μm, and a tilt angle of 70°. The TEM samples were mechanically thinned to about 50 μm and then reduced by electrolytic polishing in 10% perchloric acid at 35 V and −20°C. The SIMS samples were electro-polished and tested under 30 kV and 500 pA. Figure 3 shows OM images of the NHR alloy after solution treatments at different temperatures. Macroscopically, clean GBs including twin boundaries could be observed to contain no η phase. As shown in figure 3(a), after 980°C solution treatment, the microstructure exhibits fine equiaxed grains, implying complete recrystallization of the austenite. As the solution temperature increases, the recrystallised grains grow in size, shown in figures 3(b), (c). For a solution temperature of 1130°C, grains have grown significantly (figure 3(d)). EBSD was used to further characterise the typical microstructure, and examine the microstructural evolution during aging. Figures 4(a), (e) show EBSD images of the NHR alloy after 980°C and 1130°C solution treatments, respectively. Annealing twins were observed within the solution-treated NHR alloy, suggesting a medium stacking-fault energy for this material. The average grain size grew from 90.9 μm to 194 μm with a temperature increase from 980°C to 1130°C. The volume fraction of Σ3 boundaries was counted to be approximately 61.4% and 62.5%, respectively.

Microstructural evolution
show EBSD images of the samples solution treated at 980°C, with aging temperatures of 675°C, 725°C, 775°C, respectively. The EBSD maps present the grain boundaries of each sample are shown in Supplementary figure. S1, where the misorientation angle of the grain boundaries are highlighted in green (less than 20°), red (20°~50°) and dark (more than 50°), respectively [24]. It is interesting to note that, for the samples under 980°C solution treatment, grains continued to grow during the low-temperature aging treatments. For example, S980A675 (figure 4(b)) shows fine grains accompanied by the appearance of larger grains with sizes near 150 μm. For S980A725 ( figure 4(c)), the number of fine grains has decreased and more large grains can be  observed. For S980A775 (figure 4(d)), the remaining fine grains were further compressed and the size of large grains reaches approximately 187 μm, which is approaching the size of the grains after solution treatment at 1130°C (figure 4(e)). However, for the S1130 samples, aging does not appear to change the microstructure significantly; grains show no significant change, maintaining a size of about 200 μm at all different aging temperatures (figures 4(f)-(h)). The corresponding inverse pole figure map of NHR demonstrates a random orientation after different heat treatments.
Usually, alloy grain size does not change after a high-temperature solution treatment has been completed, especially not during or after a low-temperatue aging treatment. These EBSD results reveal abnormal grain growth in the previously fine-grained structure during low-temperature aging, which may influence the final mechanical properties of this alloy. The mechanism of this abnormal grain growth will be discussed later.
The formation of precipitates, i.e. γ′ phase, under several heat treatments was characterised by TEM.
precipitates are of the L1 2 type. Using Image-Pro Plus software, the average diameter of the γ′ phase was deterimined to be 10.43 nm, 9.88 nm, 16.28 nm, and 15.82 nm for S980A725, S1130A725, S980A775, and S1130A775, respectively. The volume fraction of the γ′ phase was determined to be 28% for both S980A725 and S1130A725, and 34% for S980A775 and S1130A775. The γ′ phase appears to have grown by about 5 nm when the aging temperature increased from 725°C to 775°C. However, grain size appears to show little influence on the growth of precipitates during aging. Using the PRISMA module in Thermo-Calc software, the kinetics of γ′ precipitate growth after isothermal aging treatments were simulated. Calculations were carried out with the database TCFE12 and MOBFE7. All elements (Fe, Cr, Ni, V, Ti, Al, Mo, Co, and W) were included for the simulation. The interfacial energy input was set at 0.1 J m −2 [25], molar volume input was set at 1.4 × 10 -6 m 3 mol -1 , morphology input was set at spheroidal, and the simulation time was 8 h for 675°C, 700°C, 725°C, 750°C , and 775°C isothermal aging treatments. As shown in figure 5(e), the calculated mean diameter agrees well with the experimental measurements (represented by star symbols). Figure 6 shows STEM-EDS elemental mapping of γ′/γ after different heat treatments. There was significant enrichment of nickel, titanium, and aluminium in small spherical particles with dispersed distribution, which further proved to be Ni 3 (Ti, Al)-type precipitates. Both figures 5 and 6 show that the characteristics of the γ′ phase are independent of solution treatment temperature. Figure 7(a) shows the strain-stress curves of NHR that has been solution treated from 980°C to 1130°C, then aged at 675°C. Figures 7(b), (c) show the curves of samples aged at 725°C and 775°C. Generally, with the increase in aging temperature, there is an increase in tensile strength and a decrease in ductility. It is shown that S980A675 has a higher work-hardening rate than S1130A675. As the aging temperature increases, the difference in the work-hardening rates of S980 and S1130 gradually decreases. When aging at 775°C, solution heat treatment temperatures do not greatly affect the tensile curves, i.e. the curves almost coincide with each other ( figure 7(c)). It is then expected that the grain refinement effect from the solution processes will have little influence on the mechanical properties when the aging temperature is over 725°C. This implies that the abnormal grain growth and the coarsening of the γ′ phase may co-result in the convergence of tensile properties.

Mechanical properties
The statistical mechanical properties of heat-treated NHR are presented in figure 8. As shown in figure 8(a), with an increase of aging temperature from 675°C to 775°C, yield strength increased approximately 66% to achieve roughly 750 MPa. Tensile strength shows a similarly increasing trend within the range of 675°C to 725°C . However, for aging temperatures above 725°C, the rate of increase of the tensile strength decreases, figure 8(b). Both yield strength and tensile strength tend to converge as aging temperatures increase, independent of the original solution heat treatment conditions. Figure 8(c) shows that the increase in aging temperatures from 675°C to 775°C decreases the ductility by about 25% and 15% for S1130 and S980 samples, respectively. The minimum ductility is still greater than 25%. Furthermore, as the aging temperature increases from 725°C to 750°C, both tensile strength and ductility show a slight increase. This suggests that 750°C could be an optimal aging temperature for the NHR alloy.
The reduction of area is directly correlated with ductility, as shown in figure 8(d). Figure 8(e) shows the decrease of impact toughness with increasing aging temperature. Because impact toughness and strength are shown to be inversely related, there is a trade-off to consider. An impact toughness of up to 40 J is possible, in exchange for relatively lower strength. To reach a strength over 1100 MPa, the impact toughness falls to roughly 30 J, for a specimen of 2.5 mm thickness.
To evaluate the general mechanical properties of the studied alloy, figure 8(f) presents the statistical results of tensile strength as a function of ductility for all samples. The NHR alloy could exhibit a combination of properties between 50% ductility and a strength of 880 MPa, or 36% ductility and strength of 1131 MPa. It is noteworthy that the solution heat treatment of 980°C failed to generate an optimal mechanical property combination. Figure 9 shows tensile fracture morphologies of NHR, each with a different treatment temperature. The fracture morphology is mainly composed of massive dimples, which is a typical characteristic of ductile fracture. In general, the dimples in the S980 samples (figures 9(a), (c), (e)) are much smaller than in the S1130 samples (figures 9(b), (d), (f)). Dimple size in S1130 samples remains constant but the dimples become increasingly shallow with increasing aging treatment temperature. When the aging temperature was raised to 775°C, a  number of cleavage facets were observed, which is widely recognised to be a feature of brittle failure [26]. This indicates that high-temperature aging can lead to a change of fracture mode from ductile rupture to brittle intergranular fracture, corresponding to the decrease in ductility and impact toughness of the NHR alloy, as seen in figure 8.

Abnormal grain growth
According to experimental results, the solution treatment temperature has little effect on mechanical properties in samples that were aged with a sufficiently high aging temperature (775°C). The abnormal grain growth during aging is the likely cause of the changing mechanical properties. In general, abnormal grain growth in other alloys is reported with an increased solution/equalization temperature. The dissolution of tiny precipitates [27] and the coarsening of second phases may both negate the pinning effect on GBs [28]. Abnormal grain growth has also been reported in relattion to the solute drag effect [29]. Generally, the solute drag effect is found to induce grain refinement in most alloys [30,31,32], by increasing the driving force for GB motion. However, a change in diffusivity may cause grains to grow under abnormal conditions. It is expected that the change in concentration of the preciptate may also release the previously hindered grain growth.
To have a better understanding of the abnormal grain growth in S980 samples during aging, we investigated features of GBs and found three implicit factors that may impede grain growth during a solution heat treatment. Firstly, we discuss the precipitate pinning effect. Figure 10 shows an STEM image and STEM-EDS elements maps of precipitates found at GBs in the NHR alloy after 980°C solution treatment. As shown in figure 10, the GB was pinned by the nickel-titanium-enriched precipitates. These particles are found to be much smaller than 10 nm, which is known to effectively impede the migration of austenitic GBs [33]. Secondly, we discuss the second phase pinning effect. In the S980 sample, η phase is not clearly found in the OM image ( figure 3(a)). However, as illustrated in figures 11(a), (b), cellular η phase could still be observed at some GBs. According to the [34], η phase tends to dissolve into the matrix at elevated temperatures, but the exact dissolution behavior may be related to local thermodynamic conditions at GBs. It is expected that the η phase can also impede the movement of the interface [35]. Figures 11(c), (d) show the dissolution/spheroidization of the η phase at GBs when aging at 725°C. This spheroidized second phase has been shown to reduce the pinning effect on moving GBs in some pearlitic steels [36].
Thirdly, we discuss the solute drag effect. As shown in figure 12(a), nickel and titanium segregate to GBs, which can be detected after a solid solution heat treatment of 980°C. It is well known that titanium tends to segregate to GBs, usually via non-equilibrium pathways. When the solution temperature increased to 1130°C ( figure 12(b)), the solubility of alloying elements is increased, making the element segregation less significant or noticable. Figure 13 highlights grain boundaries after both solution and aging heat treatments. After aging at 675°C, the segregation of nickel and titanium at GBs could still be observed, as seen in figure 13(a). This still exerts a solute drag effect and prevents the movement of GBs. After a 725°C aging treatment, there is no obvious elemental segregation at the GBs and the grains have grown significantly ( figure 13(b)).  The precipitates, η phase and elemental segregation that impede motion of GBs are shown to be mostly composed of nickel and titanium. Aging treatments lead to the formation of a coherent γ′ phase Ni 3 (Ti, Al), an ordered L1 2 structure. As a type of coherent precipitates, these decrease the free energy of the total system. The γ′ phase tends to grow in austenitic matrix and consumes the chemical content of nickel, titanium, and aluminium present in the austenitic matrix. The aging treatment allows for the dissolution of the nickel-titanium-enriched second phase and the diffusion of those elements previously segregated at GBs to meet a local equilibrium condition. Thus, the pinning effect or the drag effect on the movement of GBs will be reduced, driving GB motion. It is expected that after the formation of γ′, it will continuously coarsen, as shown in figure 5(e). It is reported that γ′ also affects grain growth via elastic misfit strain stabilization and Zener-type pinning [37]. However, at the initial stage of aging, γ′ is considered too small to impede the migration of GBs. Thus, the abnormal grain growth in the S980 samples during aging is suspected to be related to dissolution of the nickeltitanium-enriched segregation/second phase, which exhibites a pinning effect during solution heat treatments.  where G represents shear modulus, b is the Burgers vector, γ APB is the antiphase boundary (APB) energy, ω is a constant of the order of unity, d is the diameter of γ′, and f is the volume fraction of γ′. The values of G, b, γ APB , and ω are taken as 80 GPa, 0.254 nm, 0.56 J m −2 [43], and 1, respectively. According to the previous statistical calculations for the S980A725 and S1130A725 samples, d is taken as 10.43 nm and 9.88 nm respectively, and f is taken as 28%. For the S980A775 and S1130A775 samples, d is taken as 16.28 nm and 15.82 nm respectively, and f is taken as 34%.
As shown in figure 14, the dislocations will tend to move via whichever mechanism provides the least resistance to glide [44]. Consequently, WCD dominate the interactions between γ′ phase and dislocations. The strength contribution increased with the coarsening of the γ′ phase. For S980A725 and S1130A725, the contribution from precipitation strengthening was calculated as 297 MPa and 281 MPa respectively, accounting for more than 48% of the measured yield strength. For S980A775 and S1130A775, the contribution is calculated as 460 MPa and 448 MPa, accounting for more than 60% of the measured yield strength. Aging at 775°C, ductility decreased severely and a number of cleavage facets with river lines were observed in the fracture morphology. It can be seen in figure 15 that the η phase reappeared at GBs with a cellular morphology. During tensile testing, the η phase becomes the crack source, causing cleavage fracture and deterioration of mechanical properties. In figure 15, η can be observed to grow at the expense of the γ′ phase, which may account for the slowing growth in tensile strength when aging from 750°C to 775°C. Figure 16 contains a schematic to summarise the microstructural evolution during the previous heat treatments of the NHR alloy. The precipitate (rich in nickel and titanium) and the solute segregating together hinder the movement of GBs at 980°C. During aging, the γ′ phase is generated in the matrix, which consumes nickel and titanium and leads to reduced hindrance on GB motion. With increasing aging temperature, the γ′ phase grows and its volume fraction becomes larger, which further removes the hindrance on GB motion. Finally, grains grow significantly at 775°C, so that the microstructure resembles that of the sample which was solution treated at 1130°C. This abnormal grain growth phenomenon is seldom found in other alloys, which may be driving the new interest in manipulating the titanium concentration in existing alloys. According to the experimental results, grain refinement during a solution heat treatment may not be an effective way to optimise the mechanical properties of NHR. These results are expected to shed light on better post-treatment of such alloys in additive manufacturing applications. Finally, aging at 750°C may provide a better combination of strength and ductility. Two step aging, i.e. aging at a lower temperature in a second step, could be further investigated to show increasing strength by generating small precipitates.

Conclusions
This study examined the influence of solution heat treatments (from 980°C to 1130°C) and aging heat treatments (from 675°C to 775°C) on the microstructural evolution and corresponding mechanical properties of a Fe-Ni-Cr-based austenitic alloy (NHR). The following conclusions can be obtained: (1) Increased grain growth was observed with increasing solution treatment temperature from grains near 90 μm in size (980°C) to 200 μm in size (1130°C). It has been demonstrated that grain refinement could be attributed to the solute drag effect of titanium, and the pinning effect of nickel-titanium-enriched particles as Figure 15. STEM images and STEM-EDS element mapping of GB η phase in NHR after S1130A775 treatment. a second phase. Abnormal grain growth during aging was found in samples with fine austenitic structures. During aging, titanium diffused out of the GBs into the γ′ phase, thereby triggering the motion of GBs.
(2) The coarsening of the γ′ phase (Ni 3 (Ti, Al)) is related to aging temperature but almost independent of solution treatment temperature. The microstructure and mechanical properties of samples aged at 775°C show little difference as solution treatment temperature increases. Abnormal grain growth and the coarsening of the γ′ phase may co-result in the convergence of mechanical properties.
(3) The studied NHR alloy shows a wide range of mechanical properties, from a combination of a tensile strength of 1131 MPa with ductility of 36%, to a combination of a tensile strength of 880 MPa with ductility of 50%. The properties of samples solution heat treated at 980°C (fine grains) do not compare to the best mechanical properties from other treatment routes.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.