Improving Structural Stability and Thermal Stability of Copper Alloy by Introducing Completely Coherent Ceramic Dispersoids

Abstract

When ceramic particles are incoherent with copper matrices, or when large coherent strains exist due to the differences between their crystal structure and lattice parameters in traditional dispersion-strengthened copper alloys, the strengthening effect of dispersoids at high temperatures is reduced. In the present work, a Cu-0.48Al-3.5Yb₂O₃ alloy was fabricated by mechanical alloying and spark plasma sintering. The investigation results prove that completely coherent inert ceramic particle YbAlO₃ without coherent strains is introduced into the copper matrix. The microstructural evolution and thermal stability of the alloy after annealing at high temperatures are investigated and discussed, and it is found that the alloys exposed at 600~800 °C for 3 h exhibit excellent thermal stability and exceptional structural stability. The exceptional resistance to grain growth in the alloy can be attributed to the Zener pinning effect provided by the fine dispersion of YbAlO₃ particles. High-density geometrically necessary dislocation (GND) is retained in the alloy even after annealing at 800 °C for 3 h, as is the presence of parallel GND rows because they do not easily react with opposite rows to annihilate the dislocation. At the same time, dispersed YbAlO₃ acts as a strong obstacle to moving the GND. The present work proves that the structural stability of copper can be significantly improved by introducing completely coherent dispersed particles.

1. Introduction

High-strength and high-conductivity copper alloys are a potentially attractive option for a variety of demanding high-heat flux structural applications, such as heat-sink materials in fusion reactors, rocket engine combustion chambers and nozzle liners, and high-performance metal gaskets [1,2,3,4]. The main factor limiting the application of copper alloys at high temperatures is their structural stability because all current commercially available high-strength and high-conductivity copper alloys suffer from significant creep deformation at temperatures above 300~400 °C [5,6]. The rising demand for high-temperature copper alloys has attracted extensive attention in terms of improving thermal stability and mechanical properties at 400~800 °C.

It is well known that conventional strengthening mechanisms, such as cold working, precipitation hardening and grain refinement, are ineffective at high temperatures, owning to recrystallization, precipitate coarsening and grain growth, respectively. At high temperatures, unless the substructure is stabilized by dispersoids, it is altered by recovery and recrystallization processes and thus contribute little to strength. Therefore, dispersion strengthening is the most effectively used strengthening method for copper alloys at high temperatures, and a substantial amount of works have focused on particle selection for dispersion-strengthening copper alloys. Many second particles have been selected for the strengthening phase to improve the thermal stability of the copper matrix in recent works, such as Al₂O₃ [7], Y₂O₃ [8], TiC [9], WC [10,11] and Cr₂Nb [12,13,14]. In addition, refractory metal particles, such as Ta [15,16] and W [17], have also been used to improve the structural stability and mechanical properties of copper matrices at high temperatures. The investigations in the above research have proved that the thermal stability and mechanical properties of copper alloys can be improved by introducing appropriate second particles into the copper matrix. Nonetheless, it is ambiguous whether all inert ceramic dispersoids or refractory metal particles play the same role in the copper matrix in the case of consistent particle size and distribution.

Groza and Gibeling [18] established the guidelines for the particle-selection process for copper based on the interaction between the functional design requirements at high temperatures and the intrinsic material characteristics. The major particle requirements include thermodynamic and chemical stability, low solubility and low diffusivity in the copper matrix, high interfacial energy of the particle–matrix interface, and little difference in thermal expansion coefficients between the matrix and the particle. The particles in dispersion-strengthening copper alloys are usually introduced by powder metallurgy, and the crystal structure and lattice parameters between the particles and the matrix are different; therefore, the particles are always incoherent with the matrix. The dislocations move past these particles by Orowan looping in the dispersion-strengthening copper alloys because it is difficult for the dislocations to cut these incoherent and hard particles. When the dispersions possessing the same crystal structure and lattice parameters with the copper are selected as strengthening particles, can the dislocations shear these hard particles or not? Theoretically, because a coherent particle shares the same crystal lattice as its matrix, it may be shared by a dislocation moving on a matrix slip plane. When the particles are sheared by dislocations in such alloys, the strain is distributed uniformly between the particles and the matrix. This tends to minimize ductility losses that accompany strengthening in this method [19]. Moreover, the shape change and coalescence of the particles can be eliminated by alloying for zero lattice parameter mismatches according to the precipitation-strengthened alloys. This may also be suitable for dispersion-strengthened alloys.

In this study, we tried to fabricate dispersion-strengthening copper alloy with completely coherent strengthening particles. Ceramic particle YbAlO₃ with the same crystal structure as and similar lattice parameters to copper was selected as the strengthening particle. The coherency of the matrix and dispersions with different sizes was identified, and the structural stability of the copper matrix and dispersoids at 600~800 °C was investigated. The purpose of this paper is to prove that completely coherent inert particles may be more suitable for copper alloys serving in high temperatures.

2. Materials and Methods

The nominal composition of the Cu-0.48Al-3.5Yb₂O₃ (wt%) alloy was fabricated by powder metallurgy. Raw powders of Cu (99.9%, 1 μm), Al (99.9%, 1 μm) and Yb₂O₃ (99.99%, 30 nm) were milled together under a high-purity argon atmosphere in a high-energy planetary ball mill at 300 rpm for 40 h. Zirconia (ZrO₂) vials and balls were used as milling media and 3wt% ethanol was used as a process control agent (PCA). The milled powders were consolidated by spark plasma sintering (SPS) at 900 °C for 6 min under the applied pressure of 40 MPa. After sintering, the alloy was rolled at 750 °C by ~70% to obtain a flat sheet with a thickness of ~2 mm. The rolled sheet was then annealed at 600 °C for 5 h to reduce work hardening.

XRD analyses of powders and bulk samples were performed using a Bruker D8 ADVANCE X-ray diffractometer (Changsha University of Science and Technology, Changsha, China) with Cu Ka radiation. The morphology and chemical compositions of the as-milled powders were analyzed by scanning electron microscopy (SEM) with energy-dispersive spectroscopy (EDS). Grain size and geometrically necessary dislocation (GND) density of the bulk alloy were determined by electron backscatter diffraction (EBSD) in an FIB of Thermo Scientific Scios 2 with an EBSD detector (Changsha University of Science and Technology, Changsha, China). GND was obtained through ATEX software (3.28, Université de Lorraine-Metz, Lorraine, France) [20]. Microstructures of the bulk samples were obtained using an FEI Tecnai F20 (200 kV) with selected area electron diffraction (SAED) (Changsha University of Science and Technology, Changsha, China). TEM specimens were prepared by mechanical thinning followed by ion milling. A tensile test was carried out at an ambient temperature, utilizing a testing system with a strain rate of 1 × 10⁻³ s⁻¹.

3. Results and Discussion

3.1. Microstructural Evolution

Figure 1a shows SEM images of the as-milled powders, and a combination of thick flake-like and nearly ellipsoidal-shaped powders can be observed. Figure 1b shows the high-magnification morphology of a disc powder, and the corresponding EDS mappings are exhibited in Figure 1c–f. According to the EDS results, Yb, Al and O elements are uniformly distributed in the powder, which indicates that the mixed powder successfully achieves alloying.

Figure 2 exhibits XRD patterns of the as-milled powders and sintered alloys annealed at 600~800 °C. It can be observed in Figure 2a that the diffraction peaks of the annealed alloys increase to higher diffraction angles compared to those of as-milled powders, suggesting a smaller lattice parameter in the annealed alloy. The larger lattice constant of the as-milled powders can be attributed to the formation of an unstable solid solution during the mechanical alloying (MA) process due to dissolved Al and Yb₂O₃ atoms in the Cu lattice. Then, the lattice constant decreases with the precipitation of the solid solution atoms during sintering and annealing processes. The diffraction peaks of Cu and Yb₂O₃ are identified in MA powders, while Cu and YbAlO₃ peaks are confirmed in the annealed alloys, which indicates that Yb₂O₃ transforms into YbAlO₃ during the subsequent sintering, rolling and annealing processes. The formation of YbAlO₃ process is

Yb₂O₃ + 2Al + 3O → 2YbAlO₃

(1)

The formation Gibbs energy of YbAlO₃ is negative, and therefore the reaction is spontaneous at high temperatures. It has to be noted that additional oxygen atoms are required in the reaction, and additional oxygen atoms can be provided by oxygen impurity introduced from the PCA and raw powders during the MA process, which decreases the oxygen content and thus further purifies the matrix. It is difficult to individually judge the crystal structure of YbAlO₃ by XRD patterns because YbAlO₃ has several types of crystal structures, including cubic, hexagonal and orthorhombic. The nominal atomic ratio of Yb to Al is 1:1 in the Cu-0.48Al-3.5Yb₂O₃ (wt%) alloy and, theoretically, no residual Yb₂O₃ or Al exist in the matrix according to their reaction. There is no Yb₂O₃ diffraction peak detected in the annealed alloys, which suggests that most Yb₂O₃ react with Al to form YbAlO₃.

Figure 2b shows the microstrain, dislocation density and grain size of the MA powders and annealed alloys acquired from XRD patterns. The microstrain, dislocation density and grain size are obtained through following equation [21,22]:

$$\rho =2\sqrt{3}\epsilon /db$$

(2)

where ρ is the dislocation density, ε is the microstrain, d is the average grain size and b is the Burgers vector. The values of d and ε can be calculated from the XRD peak broadening (B) using the Williamson–Hall method [23]:

Bcosθ_B = K/d + ε·sinθ_B

(3)

where λ is the wavelength of Cu Kα radiation (0.154 nm), K is the constant (0.9) and θ_B is the Bragg angle. Based on the XRD patterns in Figure 2a, ε and d can be obtained.

The microstrain and dislocation density of the annealed alloys are much lower than that of the MA powders. This is caused by dislocation recovery, recrystallization and grain growth at high temperatures. The microstrain and dislocation density of the annealed alloys decreases as the annealing temperature increases because the dislocation recovery and annihilation process are positively correlated to temperature; in other words, the higher the temperature, the faster and more sufficient the dislocation reaction. Unusually, the grain size of the annealed alloy decreases slightly with an increase in annealing temperature. The only way to refine grains at high temperatures is recrystallization; therefore, the most likely explanation is that the rate of grain refinement through recrystallization is higher than that of grain growth.

The EBSD inverse pole figure (IPF) of the alloy with different annealing temperatures are shown in Figure 3a–c, and the corresponding grain size distributions are presented in Figure 3d–f. The average grain size of the alloy annealed at 600 °C, 700 °C and 800 °C was 1.25 μm, 1.24 μm and 1.31 μm, respectively. The alloys exposed to high temperatures (0.64~0.79T_m) showed negligible change in grain size, and therefore an exceptional structural stability of the alloy was demonstrated. It is well known that the Zener pinning effect provided by second particles is always the most effective way to inhibit overall grain growth, and the suppression of the grain growth in the Cu-Al-Yb₂O₃ alloy is most likely related to the presence of precipitates in the matrix. Furthermore, some subgrains and ultra-fine equiaxed grains can be observed in the annealed alloys, suggesting a partial recrystallization structure even after annealing at 800 °C for 3 h.

To estimate the crystallinity of the alloy, the recrystallization fraction (blue regions) and un-recrystallization fraction (red regions) were distinguished and are depicted in Figure 4a–c. The recrystallization fraction increased from 63% to 71% as the annealing temperature increased from 600 °C to 800 °C. With the progress of the recrystallization process, subgrain boundaries transformed into general grain boundaries, and thus the initial coarse grain was refined. At the same time, grain growth occurred during the annealing process, more or less. The decreasing fraction of ultra-fine grain (<1 μm) and coarse grain (>4 μm) as the annealing temperature increased proves that grain growth and recrystallization occur simultaneously during the annealing process. Finally, the balance between recrystallization and grain growth led to a stable grain size. Figure 4d–f exhibits the GND distribution of the alloys annealed at 600 °C, 700 °C and 800 °C, respectively. The average GND density decreased with the increase in annealing temperature from 7.2 × 10¹⁴ m⁻² at 600 °C to 6.8 × 10¹⁴ m⁻² at 700 °C, and finally decreased to 6.6 × 10¹⁴ m⁻² at 800 °C. Generally, the statistically stored dislocation (SSD) is generated during the uniform deformation in the crystal lattice, which does not induce any long-range stress; however, the GND reflect non-uniform deformation and represents the presence of the stress gradient in the crystal [24]. The high density of GND indicates non-uniform deformation and long-range stress distribution in the Cu-Al-Yb₂O₃ alloy. Heterogeneous grain and the second phase can both lead to non-uniform deformation and generate GND; however, the stress gradient in heterostructured materials can be eliminated at high temperatures, while the GND in particle-strengthened materials can be maintained at high temperatures due to the pinning effect by the second particles. Therefore, the high density of GND in the annealed Cu-Al-Yb₂O₃ alloy is related to the second particle in the matrix, which is proved by TEM analyses below.

Figure 5a shows detailed information on the relationship between recrystallization and GND. The accumulation areas of GND are distributed in un-recrystallization zones, while low-density GND is observed in recrystallization areas. It has to be noted that, regarding GND concentrates along subgrain boundaries in the un-recrystallization areas, we have reasons to believe that GND transforms into subgrain and grain boundaries as the annealing temperature increases, leading to the decrease in GND density and the increase in recrystallization fractions. Figure 5c show the density distribution of GND along the line of A to B in Figure 5b. GND density increased dramatically when crossing a subgrain boundary, and the GND density on the subgrain boundary was one order of magnitude higher than that in the grain. There must be obstacles that hinder the moving of dislocations and lead to the concentration of GND after annealing at high temperatures.

3.2. GND Characterization by TEM

Figure 6 exhibits the bright-field (BF) TEM images of the alloy annealed at 800 °C. Figure 6a shows the typical microstructure of the alloy with equiaxed grains, and most grains have a size of about 1 μm, which is consistent with the EBSD results in Figure 3. Figure 6b provides more information regarding the inside of the grains, and many dislocations can be observed in these grains. Because the orientation of the adjacent grains is different, according to the principle of dislocation extinction, some dislocations are invisible in the grains in the case of g × b = 0. To better observe the dislocations, grain A under the zone axis of [011] is characterized in Figure 6. As shown in Figure 6c, many parallel dislocation rows with different orientations are observed in the grain. It can be determined that these dislocations are GNDs, according to their morphological characteristics. GNDs are generated from the same dislocation source, so they have the same Burgers vector. As the GND on different slip planes move, they meet and react with each other, and thus forms new dislocations. Because the Burgers vector of parallel GNDs is equivalent, the reaction-generated dislocations also have the same Burgers vector, which is still characteristic of the GND. Interactions of GNDs in different directions can be observed in Figure 6c. Moreover, a wavy GND is detected in the grain, which indicates that these dislocations are pinned by obstacles during moving. Generally, the dislocation density in the annealed alloy is at a low level due to the recovery and annihilation of dislocations at high temperatures. However, most dislocations in the annealed Cu-Al-Yb₂O₃ alloy are GNDs, which are difficult to be annihilated through dislocation reactions because the meeting probability of two GND rows with opposite Burgers vectors is small. At the same time, the presence of strengthening particles acts as a strong obstacle to the moving of GNDs. Finally, high-density GNDs are retained in the Cu-Al-Yb₂O₃ alloy after annealing.

To further understand GNDs in the alloy, the selected area was imaged under two-beam dynamical conditions using three different reflections, including g = (1-11), g = (11-1) and g = (200), and the corresponding results are given in Figure 7. In each micrograph, only the reflection indicated is strongly excited. Figure 7a,b shows ordinary BF images of the annealed alloy under the zone axis of [011]. The contrast of these two images is dark because most dislocations are visible under the zone axis of [011]. A dislocation band composed of parallel GND is observed in Figure 7b, and the distance between each GND line is very small. Some dislocation lines are even connected in some regions. When the parallel dislocation rows slide on the slip plane, the distance between the leading dislocation and trailing dislocation becomes smaller and smaller when the leading dislocation line is pinned. In Figure 7b, a strong particle with a size of about 100 nm at the grain boundary is found as an obstacle to GND moving. With the continuous moving of dislocations, dislocation walls gradually form, and a subgrain boundary is finally built. To a great extent, the concentration of GNDs characterized by EBSD in Figure 4 can be attributed to this reason. Figure 7c,e,f show BF images of the alloy annealed at 800 °C using reflections of g = (11-1), g = (1-11) and g = (200), respectively. Parallel dislocations along the g = (11-1) direction are visible in Figure 7c,e while they are invisible in Figure 7f, which indicates that the Burgers vector of these dislocations is b = 1/2[01-1]. Many short dislocation lines along the g = (1-11) direction are visible in Figure 7e,f, while they are invisible in Figure 7c, and thus b = 1/6[112] is identified as their Burgers vector. According to the distribution of perfect dislocation b = 1/2[01-1] and partial dislocation b = 1/6[112] on the Thompson tetrahedron, they will not react with each other to annihilate the dislocation, so these dislocations are retained even after annealing at 800 °C. Similar dislocation structures can be found in each grain. This is the reason why high-density dislocations are preserved in the alloy after annealing at high temperatures.

3.3. Nanoparticle Investigation

Figure 8a shows a BF image of the alloy, and some nanoparticles marked by blue circles can be observed in the grain. It has to be noted that when the nanoparticles possess the same crystal structure and lattice parameters as the Cu and are completely coherent with the matrix, the strain at their interface is very small, even close to zero, which may make the nanoparticles difficult to observe. Therefore, the nanoparticles are directly observed in high-resolution TEM (HRTEM) images in this work. Figure 8b,c exhibits two nanoparticles with diameters of about 9 nm and 30 nm, respectively. In Figure 8b, the interspacing of the matrix and the particle in two directions are measured as 0.208 nm and 0.181 nm, corresponding to d(111) and d(200) of Cu and YbAlO₃, respectively, which confirms that YbAlO₃ with same crystal structure and lattice parameters as Cu was successfully introduced into the matrix. To confirm the coherency of the interface between the nanoparticles and their surrounding matrix, IFFT was conducted on planes of (1-11), (200) and (11-1), and the corresponding results are given in Figure 8d–i, respectively. No lattice misfit was observed around these two nanoparticles, suggesting completely coherent interfaces between the matrix and the particles. In previous research about dispersion-strengthening copper alloys, large coherent stain existed around the dispersoids when they were coherent or semi-coherent with the matrix due to the differences in their crystal structure and lattice parameters. Even stable coherent particles may not yield effective strengthening at high temperatures if there are large coherency strains that lead to rapid coarsening by shape change and coalescence, although these strains can improve the strength of the alloy at room temperature through interaction with dislocations. In this work, no coherent strain or lattice misfit were observed around the coherent dispersoids with different sizes; therefore, these nanoparticles may have high thermal stability and coarsening resistance.

3.4. Thermal Stability and Mechanical Property

To estimate the thermal stability of the alloy, the hardness of the sintered and annealed alloys was tested and the results are given in Figure 9a. The hardness values of the sintered and annealed alloys were all about 130 HV. The hardness of the alloy was still stable after annealing at 600~800 °C for 3 h, demonstrating the excellent thermal stability of the alloy even at 800 °C. The superior thermal stability of the alloy can be attributed to its exceptional structural stability, including extremely sluggish grain growth and coarsening resistance of dispersoids. Figure 9b exhibits the engineering stress–strain curve of the alloy annealed at 800 °C for 3 h, in which the tensile strength and elongation of the alloy were obtained at 377 MPa and 9%, respectively. The strength of the alloy is mainly contributed by grain boundary strengthening, dislocation strengthening and dispersion strengthening, according to the experimental results in this work. However, it is not clear whether the coherent dispersoids are sheared or bypassed by dislocations during the deformation process. Further work needs to be conducted to confirm the interactions between the dislocations and these completely coherent dispersoids. If dispersoids are sheared by dislocations, the strain is distributed uniformly between the matrix and nanoparticles. This tends to minimize ductility losses accompanying strengthening by this method. Figure 9c,d shows the tensile fracture morphologies of the alloy annealed at 800 °C. Numerous dimples with different diameters were observed on the fracture surfaces, which suggests a ductile fracture of the alloy.

4. Conclusions

In the present research, the Cu-0.48Al-3.5Yb₂O₃ alloy with extremely high thermal stability and structural stability was fabricated through mechanical alloying and SPS. The mechanical properties and microstructural evolution of the alloy after annealing at 600 °C~800 °C were characterized, and the main conclusions are listed as follows:

(1) YbAlO₃ possessing the same crystal structure and lattice parameters as Cu was successfully introduced into Cu matrix after sintering by the reaction between Yb₂O₃ and Al. These YbAlO₃ particles were completely coherent with the copper matrix and no coherent strain was observed around the particles.

(2) The alloys exposed at 600 °C~800 °C for 3 h exhibited negligible grain growth and stable hardness. The exceptional structural stability and thermal stability of the alloy were contributed by the Zener pinning effect provided by YbAlO₃ particles. The tensile strength of the alloy after annealing at 800 °C for 3 h was 377 MPa, and the strength was mainly contributed by grain boundary strengthening, GND strengthening and dispersion strengthening.

(3) High-density GND was retained in the alloy even after annealing at 800 °C. The parallel GND rows were not easy to annihilate through the dislocation reaction at high temperatures. The presence of YbAlO₃ particles also acted as obstacles to the inhibition of the moving of GND, leading to concentrations of GND in the grain. The dislocation pileup will gradually form a dislocation wall with the moving of parallel GNDs, and finally transform into the subgrain boundary.