The Effect of Thermomechanical Processing Sequence on the Dispersoid Distribution and Final Mechanical Properties of Spray-Formed Al-Cu-Li Alloy

Abstract

Controlling the formation of the β′ (Al₃Zr) phase is pivotal for regulating the recrystallization and thus the mechanical properties of the spray formed 2195 (Al-Cu-Li) alloy. In a conventional “homogenization-extrusion” process, the precipitation of β′ is severely affected by the presence of the T₁(Al₂CuLi) phase in the as-deposited alloy, leading to an inhomogeneous distribution of the β′ phase. In the present work, we propose a new thermomechanical processing (TMP)—swapping the order of the homogenization and extrusion processes. The microstructures and properties of the new proposed TMP were systematically studied at various stages of the alloy treatment and compared with the out of the conventional TMP. It was revealed that the introduction of the extrusion process on the as-deposited alloy can break the continuous network of primary phases and dissolve the T₁ phase, promoting a uniform distribution of the β′ phase during subsequent two-step homogenization. During solution treatment, the new TMP is more effectively in suppressing the formation of a coarse grain layer at sheet surface, while after final peak aging, the new TMP produces a lower alloy strength but a higher elongation, due mainly to the smaller thickness reduction during deformation. The new proposed TMP technique provides a new insight into regulating the mechanical properties of Al-Cu-Li alloys.

1. Introduction

Al-Li alloys have received increasing attention in recent years because of a good combination of excellent properties, such as low density, high specific strength, and high corrosion resistance. As a typical third-generation Al-Li alloy, the 2195 Al-Cu-Li alloy has been widely applied in aerospace and military fields in the past decades owing to its low anisotropy, high thermal stability, and good fatigue resistance [1,2]. Intensive efforts have been made to further improve the mechanical properties of this alloy. Spray forming, a typical rapid solidification technology, has emerged as a promising candidate for manufacturing high-performance aluminum alloys [3,4,5,6]. Compared with the traditional casting process, spray forming can significantly reduce microsegregation, refine the grain size and change the size, morphology, and distribution of the second-phase particles, thereby significantly improving the mechanical properties of the 2195 alloy [7].

Regulation of thermomechanical processing (TMP), including homogenization, extrusion, rolling, solution treatment, and artificial aging, is the key issue for controlling the mechanical properties of the spray-formed 2195 alloy [7,8,9,10]. Homogenization is the initial and indispensable process for the as-cast alloy, required for the elimination of solute segregation, and facilitating subsequent rolling/extrusion processing [11,12]. Liu et al. [6] studied the microstructural evolution of spray-formed Al-Cu-Li alloy during homogenization, they found that during homogenization at 500 °C, various kinds of primary phases, including TB (Al₇Cu₄Li), θ(Al₂Cu), R(Al₅CuLi₃) and S(Al₂CuMg) phases are completely dissolved into the matrix, the dendritic segregation is eliminated, while the insoluble Al₇Cu₂Fe constitute phase is almost no change during the homogenization. Besides, for Al-Cu-Li alloy containing Zr, precipitation of metastable L1₂-Al₃Zr (β′) dispersoids occurred during homogenization, these Al₃Zr particles could not only provide the Orowan strengthening, but also inhibit recrystallization during the subsequent thermo-mechanical process and hence retaining the fibrous grains, which improve the mechanical properties of the alloy.

The effectiveness of dispersoids in inhibiting recrystallization depends on their sizes, spacings, density, and distributions. However, due to the microsegregation of Zr atoms during solidification, the distribution of the β′ dispersoids across a grain is usually inhomogeneous, which affects the final properties of the alloys [13,14]. Especially, the occurrence of a coarse recrystallized grain layer at the surface of the sheet can severely degrade the properties of the alloys. It has been reported [6,15,16] that the distribution of dispersoids can be effectively regulated by optimizing the homogenization processes, such as lowering ramp heating rate, two-step homogenization, etc. Xu et al. [17] found that the smaller heating rate (30 °C /h) can obviously improve the dispersoids distribution and hence the recrystallization resistance. Deng et al. [11] and Liu et al. [6] reported that, compared with single-stage homogenization, the two-stage homogenization treatment produces a higher density and more uniform distribution of β′ phase, which significantly suppresses the recrystallization and improves the properties of the alloy. For spray-forming Al-Cu-Li alloys, controlling the distribution of β′ dispersoids is even more challenging. As the spray-forming involves rapid solidification of the melt during droplet spray, while relatively slow cooling during the deposition, resulting in the precipitation of a high density of T₁ phase in the deposited alloy. The presence of the T₁ phase can promote the heterogeneous nucleation of the β′ dispersoids during homogenization, causing the inhomogeneous distribution of the β′ particles [16]. Besides, the relatively low temperature of the droplets reduces the solubility of the Zr atoms, resulting in low nucleation drive force of the β′ dispersoids. Therefore, there is an urgent demand for regulating the formation of β′ dispersoids by designing a novel homogenization process.

In the present study, we proposed a new thermomechanical processing (TMP)—switching the sequence of homogenization and extruding processes. The new TMP technique can effectively resolve the inhomogeneous distribution of the β′ dispersoids during homogenization. By applying the new TMP, although the final yield strength is slightly decreased, the elongation of the alloy is significantly improved. The development of the new TMP technique provides a new insight into regulating the mechanical properties of Al-Cu-Li alloys.

2. Materials and Methods

The current study was carried out on a spray formed 2195 alloy with the composition of Al-3.9Cu-0.93Li-0.52Mg-0.35Ag-0.11Zr (wt.%). The large-scale spray formed 2195 alloy billet with Φ600 mm in diameter was provided by the Haoran Co., Ltd. Jiangsu, China.

Two different thermo-mechanical treatments (TMT) were carried out for the ingot. (1) Process one (P1): homogenization–extrusion. The ingot was firstly two-stage, ramp heating homogenized at 350 °C/10 h + 470 °C/24 h (the heating rate of 50 °C/h) and water quenched to room temperature. Then hot extrusion at 470 °C with a ratio of 9:1 and a speed of 1.8 mm/s was carried on homogenized ingot for eliminating the porosity. (2) Process two (P2): extrusion-homogenization, the sequence of extrusion and homogenization is changed compared with P1, the parameters of the extrusion and homogenization are the same as P1. After then, the P1 and P2 processed alloys were hot rolled at 470 °C with a total reduction of 60%. The sheet was then solution treated at 510 °C/1 h, water quenched to room temperature and artificially aged at 170 °C for 12 h to obtain peak aging condition. The peak aged condition is selected based on the previous literature [7]. The schematic illustration of the two thermo-mechanical treatments is shown in Figure 1.

The grain morphology and distribution were characterized by polarized light optical microscope (PLOM) on ZEISS-AXio Observer 3 m (Carl Zeiss Microscopy Gmb, Jena, Germany). The size and distribution of the β′ phase were observed by an FEI Tecnai G2 F20 transmission electron microscope (TEM) (FEI, Portland, OR, USA) operated at 200 KV. The microstructure of the alloys under different processes was conducted on a Zeiss Auriga focused ion bea-scanning electron microscope (FIB-SEM) dual-beam system (Carl Zeiss Microscopy Gmb, Jena, Germany) operated at 20 kV. For microstructure and orientation analysis, electron back-scattered diffraction (EBSD) measurements were performed in a Zeiss Auriga FIB-SEM with a probe current of 20 nA and a scanning step size of 1 μm. Data acquisition and analyses were carried out using the TSL OIM™ Analysis software, version 7 (EDAX, Mahwah, NJ, USA). Cleanup with a grain tolerance of 2° was applied to re-indexing the data point before the data processing. For coarse grain with an irregular shape, the effective grain diameter R₀ was defined as:

$${\mathrm{R}}_0 = 2\sqrt{\frac{\mathrm{A}}{\mathsf{\pi }}}$$

(1)

A represents the area of one grain. The nearest is set to 1st and the maximum misorientation is 5° for the calculation of kernel average misorientation (KAM). Recrystallized area is defined by the misorientation below 2°, subgrain is defined by the misorientation between 2 and 7°, and deformed area is defined by the misorientation above 7°. The tensile test was carried out on an MTS C43.504 type testing machine (MTS, Eden Prairie, MN, USA ) according to the national standard GB/T 228.1-2010. Samples with a 25 mm gauge along the rolling direction were stretched at a rate of 1.5 mm/min.

3. Results

3.1. Microstructure of Sprayed Condition

Figure 2 shows the representative SEM image and energy dispersive spectroscopy (EDS) quantification results of the as-deposited alloy. As shown in Figure 2a, the as-deposited alloy consists of fine equiaxed grains with grain sizes of about 30–50 µm. Some pores are visible in the alloy, which is the main drawback of the spray-formed alloy. Some discontinuously or semi-continuously network primary phases are distributed along the grain boundaries. Besides, abundant plate-like and irregular morphology phases are observed in the grain interior. Figure 2b shows the SEM image of primary phases under higher magnification. The compositions of these second phases were analyzed by EDS, and the results are shown in Figure 2c. Various types of primary phases, including Al-Cu, Al-Cu-Fe, Al-Cu-Mg, and Al-Cu-Zr phase can be identified. By comparing with the results reported in literatures [16,17,18], it can be judged that the Al-Cu phase is θ′(Al₂Cu), the Mg-containing phase is S(Al₂CuMg), and the Fe-containing phase is Aplate-likee plate -like particles in the interior of the grains can be identified as the T₁ phase according to previous literature [16].

3.2. Microstructure of As-Homogenized Conditions

Figure 3 shows SEM images of the as-homogenized alloy (P1 process). After the introduction of two-stage homogenization, most of the primary phases dissolve into the matrix. The fine plate-like T₁ phase in the interior of the grain completely disappeared. The network phases along the grain boundaries are also largely dissolved, but some needle or irregular-like residual phases are remaining in the matrix. The enlarged SEM image and corresponding EDS analysis of the residual phases are presented in Figure 3b,c. Two types of residual phases can be identified: Al₇Cu₂Fe and AlCuZr phases. Most of the AlCuZr phase is closely connected with the Al₇Cu₂Fe phase. These two phases are highly stable [19] and exhibit nearly no change in size, morphology, and content after homogenization.

From the high-resolution SEM images shown in Figure 3d,e, a high density of fine spherical particles with sizes of about 50 nm is observed, these spherical particles can be identified as the β′ (Al₃Zr) phase in terms of the published literatures [6,12,20]. Interestingly, in addition to the diffuse distribution of the spherical β′ phase, most of the β′ particles are aligned in the straight chains, displaying two different orientations with a crossover angle of 70.5°, corresponding to the orientations of the T₁ phase observed in the as-deposited alloy (equivalent {111}_Al planes). As a result of these observations, it can be deduced that the β′ phase is formed at locations where the T₁ phase has dissolved. Similar behavior has also been reported by Wang et al. [16]. In addition, some shorter and curved chains of β′ precipitates are also observed in Figure 3e. These chains are not obviously aligned with any pre-existing precipitates, but are likely formed on matrix dislocations due to the reduced strain energy barrier in this region.

SEM images of the extruded alloy are shown in Figure 4a. Compared with the as-deposited alloy, it can be seen that the continuous network primary phases along the grain boundaries are broken and all the second phases are uniformly distributed in the alloy, the plate-like T₁ phase also disappears in the alloy. In addition, all these particles exhibit an elongated morphology with their long axis parallel to the extrusion direction. The enlarged SEM image and corresponding EDS mapping of the second phases are shown in Figure 4b–e. Three main primary phases can be identified: Al₂Cu, AlCuZr, and Al₇Cu₂Fe phases. This result indicates that although the sizes, morphologies, and distributions of the primary phases are changed after extrusion, the types of the primary phases are unchanged, which is mainly due to the insufficient heating (470 °C) holding time of the extrusion process.

After the introduction of two-stage homogenization, most of the primary phases are dissolved, and some residual phases remain in the Al matrix as shown in Figure 5a. This phenomenon is similar to process P1. EDS mappings of the two regions marked by red and blue rectangles are shown in Figure 5b. The coarse rod-like particles delimited by red rectangles are identified as Al₇Cu₂Fe phase, while the two irregular dot-like particles are identified as AlCuZr phase. These two residual phases can also be observed in P1. However, the phenomenon of β′ phase precipitated as chains in P1 does not exist.

To reveal the distribution of the β′ phase, the TEM bright-field image and corresponding selected area diffraction pattern (SAED) of the spherical particles taken along [112]_Al are shown in Figure 5d,e. The presence of additional diffraction spots at the center of the Al spots confirms the L1₂ structure of the β′ phase. It can be seen that the β′ particles are uniformly distributed in the Al matrix, and no linear array of β′ phase is observed in the alloy, suggesting the inhomogeneous distribution of the β′ phase observed in the P1 process is suppressed in the P2 process. Figure 5e shows the distribution of β′ phase close to the grain boundary. It can be seen that a dispersoid-free zone with a thickness of 600–800 nm is formed close to the grain boundary.

3.3. Microstructures Evolution after Solution Treatment

The P1 extruded and P2 homogenized samples were hot rolled and solution treated at 510 °C for 2 h. SEM micrographs of the P1 and P2 alloy after solution treatment are shown in Figure 6. Coarse elongated second-phase particles are distributed with their long axis parallel to the rolling direction. These particles can also be identified as Al₇Cu₂Fe and AlCuZr phases, indicating these two phases are stable during the whole TMP processing. The second phase distributions in the two alloys are almost the same after solution treatment.

To reveal the effect of β′ phase in inhibiting the recrystallization of the two alloys after solution treatment. Polarized light optical microscopy (PLOM) at the longitudinal cross-section (RD-ND) of the two alloys is shown in Figure 7. Both alloys exhibit mainly a fibrous microstructure. Some necklace-like tiny grains can be observed at the fiber interface, indicating partial recrystallization occurred during solution treatment. However, on the surface of the two alloys, an abnormally coarse grain layer appears. The coarse grain layers in the P1 and P2 alloys are measured as 100 and 50 μm, respectively, indicating that the P2 process can obviously suppress the surface recrystallization of the alloy during solution treatment compared with the P1 process.

In order to examine the effect of Al₃Zr precipitation on the recrystallization resistance of 2195 alloys, EBSD experiments were performed after solution treatment, as shown in Figure 8. Figure 8a–d shows the inverse pole figure (IPF, viewed along TD direction) and corresponding grain spread (GOS) maps superimposed with high and low-angle grain boundaries of the P1 and P2 alloys. From Figure 8a,b, the two alloys exhibit slender fiber structure with some tiny recrystallized grains formed at the interface of the fiber grains. From the GOS map shown in Figure 8c,d, the microstructure of the P1 alloy is characterized by a small number of recrystallized grains (about 24.2%). While in the P2 alloy, the recrystallized grains increase to 26.1%. This can further be verified from kernel average misorientation (KAM) in Figure 8h, where the misorientation between a grain at the center of the kernel and all points at the perimeter of the kernel are measured [21]. It is clear that the number fraction of KAM is higher for the P2 alloy. The average grain sizes are measured as 3.14μm for the P1 alloy and 6.05 μm for the P2 alloy (Figure 8e,f), respectively, indicating that the P1 process produces smaller grain size compared with the P2, which is beneficial for the strength of the alloy. Figure 8g shows the misorientation angle distribution (correlated), and the blue dashed line represents the theoretical distribution for a randomly oriented assembly of grains according to the results reported by Mackenzie [22], it is clear that both the two alloys exhibit nearly random distribution. The misorientation angle distribution was defined for low-angle grain boundaries (LAGBs, 2° < θ < 10°) and high-angle grain boundaries (HAGBs, θ ≥10°), which are marked with red and black lines respectively, as shown in Figure 8c,d. The fractions of HAGB (θ ≥ 10^o) are calculated as 52.9% and 56.3% for P1 and P2 alloys, respectively, which further indicates a relatively higher fraction of recrystallized grains in the P2 alloy.

3.4. Mechanical Properties of Solution Treatment Sheet

Figure 9 shows the tensile strength, yield strength, and elongation of P1 and P2 alloys under peak aged condition. It is evident that the yield, tensile strengths of the P1 alloy (566.0 $\pm$ 10.8 MPa, 605.8 $\pm$ 10.5 MPa) are higher than those of the P2 alloy (531.1 $\pm$ 5.8 MPa, 573.1 $\pm$ 12.9 MPa), while the elongation of the P1 alloy (12.7 $\pm$ 0.7%) is lower than the P2 alloy (14.4 $\pm$ 1.2%).

4. Discussions

The above experimental results demonstrated that switching the sequence of homogenization and extrusion can to some extent affect the microstructure and mechanical properties of the spray formed 2195 alloy, especially the distribution of the β′ phase. Firstly, the two-step, ramp heating homogenization is effective in promoting the precipitation of β′ phase. During the low-temperature stage, the rather high nucleation rate and low growth rate (limited by the slow diffusion of Zr element in the Al matrix) of the β′ phase lead to the formation of a large number of tiny β′ nuclei. During the second high-temperature homogenization, the rather high diffusion rate of Zr can promote the growth of the β′ nucleus, leading to the formation of a high density of β′ particles [6,23]. Similar two-step homogenization processes have been reported in previous literatures [12,23,24]. The formation of dense β′ particles during two-step homogenization plays a central role in controlling the recrystallization of the alloy.

For the P1 process, due to the high density of T₁ phase being formed during spray forming, it was found that most of the β′ particles tend to precipitate at the positions of the T₁ phase, despite its dissolving finally, leading to string-like distribution of the β′ phase. This result indicates that the T₁ phase can promote the precipitation of the β′ phase. The main reason for this phenomenon is the block effect of the T₁ phase on dislocation motion. The high density of dislocations accumulates on the interface of the T₁ plates, the Zr atoms may preferentially diffuse and form β′ phase at T₁ sites due to the high strain energy involved in the dislocations [17]. In addition, there are also some short and curved segregated β′ phases in the grains, indicating β′ phases can directly nucleate on the dislocations. The inhomogeneous distribution of the β′ phase is detrimental to the properties of the alloy. For the P2 process, the applied extrusion on the as-deposited alloy can break the continuous coarse primary phases, and all the plate-like T₁ particles are dissolved in the matrix. During subsequent two-step homogenization, the formation of β′ phase is no longer influenced by the T₁ particles. The heterogeneous nucleation of the β′ phase is suppressed and thus most of the β′ particles are uniformly distributed in the Al matrix.

The different distributions of the β′ phase formed upon the two processes produce different microstructures and thus different mechanical properties. It had been known that the coarse grain layer at the sheet surface is mainly caused by deformation inhomogeneity during the rolling [9]. Normally the grains at the surface experience large deformation levels and thus higher stored energy compared with the grains in the center. Therefore, these highly deformed grains are readily recrystallizing during solution treatment and thus produce a coarse grain layer at the surface. The formation of a high density of β′ phase during two-step homogenization can impose a dragging force on grain boundary migration by the classical Zener pinning theory, suppressing the recrystallization at the sheet surface [25]. To confirm the effect of β′ phase in suppressing the recrystallization of the sheet, another solution-treated 2195 alloy without homogenization treatment was performed, in which few β′ phases are detected in the alloy [15]. The IPF map from the edge to the center is shown in Figure 10, it is obvious that a rather thick coarse grain layer with a thickness of about 500–700 μm is formed at the sheet surface, which is much thicker than the P1 and P2 processes. The uniformly distributed of β′ phase formed in the P2 process is more effective in suppressing the formation of a coarse grain layer compared with the inhomogeneous distribution of β′ phase formed in the P1 process. Therefore, the P2 alloy has a smaller coarse grain layer compared with the P1 alloy.

For the mechanical properties of the two processed alloys after peak aging, various strengthening mechanisms including solute atoms, grain boundaries, dislocations, and nano-precipitates are responsible for the strengthening of the two alloys. As the P1 and P2 processed alloys have the same composition and undergo the same solution and aging treatments, the solute strengthening and precipitation strengthening should be similar [2,26], and the main reasons responsible for the different properties of the two alloys are grain boundary strengthening and dislocation strengthening. As shown in Figure 8e,f, the P1 alloy has smaller grain sizes but a higher density of dislocations than the P2 alloy. This phenomenon is mainly due to the different thickness reductions of the two alloys. As the extrusion is applied ahead of homogenization for the P2 alloy, the alloy is fully recrystallized after the homogenization, and thus the thickness reduction of the P2 alloy mainly occurred during hot rolling. Meanwhile, for the P1 alloy, the alloy undergoes thickness reduction during continuous extrusion and hot rolling processes, which produces a higher deformation level compared with the P2 alloy. Therefore, the slightly larger grain size but low dislocation density of the P2 alloy result in a low strength but high elongation. The present work demonstrated that switching the sequence of extrusion and homogenization can not only effectively suppress the coarse grain layer at the sheet surface, but also produces lower strength but higher elongation for the spray formed 2195 alloy. This result is beneficial for developing new thermomechanical processing for various aluminum alloys. In order to further optimize the new thermomechanical treatment, multi-phase-field simulations [27,28] of the microstructural evolution of the P1 and P2 processed alloys will be conducted in our further work.

5. Conclusions

In summary, this work studied the effect of two different thermo-mechanical treatments on the microstructures and mechanical properties of the spray formed 2195 alloy by various techniques, including PLOM, SEM, EBSD, TEM and mechanical tests. The main conclusions were drawn as follows:

• Switching the order of homogenization and extrusion will affect the microstructure and mechanical properties of the spray formed 2195 alloy. Direct two-step homogenization on the as-deposited alloy results in an inhomogeneous distribution of the β′ phase due to the preferential nucleation of β′ phase on the T₁ plates. The introduction of extrusion on the as-deposited alloy can obviously break the continuous coarse primary phases and dissolve the T₁ phase, and during the following two-step homogenization, most of the β′ particles are uniformly distributed due to the absence of the T₁ phase.
• The refined and uniform distribution of the β′ phase in the “extrusion-homogenization” process produces a smaller coarse grain layer in the sheet surface compared with the “homogenization-extrusion” processed sample with inhomogeneous distribution of the β′ phase.
• Compared with a direct homogenized sample, the “extrusion-homogenization” process produces larger grain sizes and low density of dislocations due to the low thickness reduction during the deformation. Therefore, switching the sequence of extrusion and homogenization produces lower strength but higher elongation for the spray formed 2195 alloy after peak aging.