More than 50 samples have been studied across the whole concentration and temperature range of Al-Cu phase diagram. Combining the results from DSC, SEM-EDX and XRD allows a complete description of the phase diagram to be proposed; shown in Figure 2. The results are mostly in very good qualitative agreement with the phase diagram of Ponweiser et al. But the focus of the current study was on those parts of the phase diagram which have not been described satisfactorily previously. Furthermore, phase equilibria at 300 °C have been investigated.
Solidification of the θ-Phase
A calculated phase diagram of the Al-Cu system was published by Liang and Schmid-Fetzer where a semi-empirical CALPHAD-type assessment of the system was presented. The peritectic reaction L + η → θ was modeled but it was found that the calculated liquidus line in the composition range 33 at. pct < x(Cu) < 45 at. pct did not agree well with the experimental results published by Ponweiser et al. A new CALPHAD-type assessment attempts to provide better agreement between calculation and experiment for the liquidus in the 33 at. pct < x(Cu) < 45 at. pct concentration range, but this required a change in the nature of the solidification behavior of the θ-phase from peritectic to congruent.
Congruent melting has been proposed previously,[10,11] but Goedecke suggested that this reaction was in fact metastable. Because of these disagreements, the concentration range between 31 and 40 at. pct Cu has been reinvestigated in the present work. The solidification behavior of the θ-phase was studied using samples that had been melted, homogenized and then quickly cooled in the arc furnace. Two sets of samples were prepared with compositions close to 33.3 at. pct Cu; a set with slightly higher and a set with slightly lower copper contents. The morphology of the as-cast samples was studied using SEM. A eutectic microstructure was observed in both sets of samples, but the morphology of the microstructures was different for each set of samples (see Figure 3(a) for the Al-rich side and Figure 3(b) for the Cu-rich sample) as would be expected for a congruent reaction.
In the second step, the melting behavior of the θ-phase was studied under conditions close to equilibrium and hence to the phase diagram for the thermodynamically stable system. Several samples with nominal composition between 32% Cu and 40% Cu were long-term annealed at 540 °C or 582 °C. The temperatures of the phase transitions were measured by DSC using sample cooling and heating rates of 1, 5 and 10 °C min−1, respectively. The data are presented in Figure 4, which also shows an enlarged portion of the phase diagram in the vicinity of the θ-phase. For all heating rates, both single phase samples containing the θ-phase, as well as samples exhibiting a two-phase morphology (θ + η′), indicate an invariant reaction at 589 °C, which is the temperature of the peritectic reaction L + η ↔ θ. There is no evidence of congruent melting in the long-term annealed samples.
Therefore, congruent melting of the θ-phase was confirmed to be a metastable reaction and the observed microstructures were found to be in very good agreement with the metastable version of the phase diagram published by Goedecke. The discrepancy between the calculated and experimental liquidus line in this region is probably due to the fact that the metastable reactions were observed during DSC studies of Ponweiser et al.
It is worth discussing the observed thermal effects associated with the liquidus that were seen at different heating rates in more detail. While the temperatures of the invariant reactions are independent of the heating rate, the (monovariant) thermal effects associated with the liquidus strongly depend on the heating rate (Figure 4). This is due to the fact that the liquidus thermal effect is superimposed on the peak maximum associated with the invariant when the two temperatures are very close to one another. In such a case, the thermal effect of the liquidus can no longer be separated and evaluated from the DSC curves. At higher heating rates, the peak maximum from the invariant is shifted to higher temperatures. Consequently, the lowest heating rate (1 °C min−1) yields the best representation of the true liquidus temperature and the liquidus line was drawn accordingly in Figure 4.
The samples that were annealed at 540 °C, 550 °C, 582 °C were studied to determine the location of the Cu-rich phase boundary of the θ-phase. The position of this phase boundary is temperature independent and lies at 33.5 at. pct Cu. The solubility limit of Cu in the η′-phase is 51.5 at. pct at 550 °C and 51.8 at. pct at 582 °C. The X-ray powder diffraction pattern of the Al-41.9 at. pct Cu sample that was annealed at 550 °C is shown in Figure 5. Lattice parameters that were obtained by Rietveld refinement were a = 12.1066(3) Å, b = 4.1024(1) Å, c = 6.9213(2) Å and β = 54.994(2)° for the η′-phase, and a = 6.0603(1) Å, c = 4.8763(1) Å for the θ-phase, respectively.
The ε′ + γ′ Two-Phase Field
The phase boundaries in the high-temperature (ε′ + γ′) two-phase field have not been described satisfactorily in the literature. Several samples with nominal compositions close to Al-60Cu were long-term annealed at temperatures of 688 °C, 710 °C and 750 °C to determine these compositions. A selected micrograph, taken with an SEM in BSE mode, of the sample with the overall composition Al-60.1Cu that had been annealed at 710 °C, is shown in Figure 6. According to the present results, the (ε′ + γ′)/γ′ boundary is located between 61% Cu at 682.1 °C and 63.3% Cu at 846 °C (see Table IV, reactions number 7 and 9).
The γ′ to δ Phase Transition
Most of the previous studies of this part of the Al-Cu system agree on the existence of two separate phase fields for the δ and γ′-phases. The γ′-phase is cubic and crystallizes with the Al4Cu9-type structure, while the δ-phase can be described by a rhombohedrally distorted superstructure of the Al4Cu9 type. However, the location of the phase boundaries separating the two different phase fields have not been fixed reliably in any previous study. Ponweiser et al. investigated this region by analyzing (by SEM/EDX) a fine raster of samples that had been annealed at 450 °C and found that all were single phase. Powder XRD analyses revealed that compositions between 60 and 63 at. pct Cu contained the single δ-phase. At 64 at. pct Cu, the diffraction pattern could not be evaluated completely, but in samples with higher Cu-contents the cubic γ′-phase was found.
In the current study, several samples with different compositions were annealed at different temperatures to locate the boundaries of the (δ + γ′) two-phase field. However, it was not possible to determine the phase boundaries of δ and γ′-phase fields by SEM-EDX measurement of the phase composition. The compositions of all grains in all samples were in agreement with the overall sample composition within the experimental error of the measurement. Thus, all samples were single phase based on EDX characterization, which is in complete agreement with. On the other hand, careful analysis of the X-ray powder patterns showed that only two of the samples could be refined completely with one single structural model: the single δ-phase was identified in the sample Al-61.8Cu (615_7) and the single γ′-phase in the sample Al-67.1Cu (750_8). These two samples are marked with a circled star in Figure 7. All other samples in the composition range x(Cu)=(62-67) have XRD patterns that were evaluated as a combination of both γ′ and δ-phases even for a temperature higher than the presumed upper temperature limit of the δ-phase (682 °C). Samples give a diffraction pattern corresponding to either the δ or the γ′-phase but with small additional peaks which could not always be fitted well with the applied structural models.
The discrepancy between the SEM and XRD results may be explained by the fact that most of the samples investigated undergo the γ′ → δ phase transition during quenching. They were single phase γ′ at the temperature of annealing but were forced to undergo a fast rhombohedral distortion during the quenching process. Such a transition is usually incomplete as the structure may distort locally, but diffusion is not fast enough to allow macroscopic phase separation. Thus, it is proposed that all samples located at compositions between the two dashed lines in Figure 7 were frozen in a state of structural transition between γ′ and δ during the quenching process. The same is true for the sample at 64 at. pct Cu that had been annealed at 450 °C as reported in Reference 1. Using this hypothesis, it was possible to propose the location of the equilibrium two-phase field (γ′ + δ) as shown in Figures 2 and 7. It is strongly temperature dependent and shows a significant widening of the δ single phase field at lower temperatures.
High-Temperature Phase Transition Between γ′ and γ
The γ-phase is formed by the peritectic reaction L + β ↔ γ at 991 °C (Table IV, reaction 3). Ponweiser et al. defined the γ′/γ phase transition as second order because they did not observe any indication of an invariant reaction involving the γ and γ′-phases. This conclusion was based on DTA measurements and on earlier results that were published by Liu et al. The temperature stability of the γ′-phase has been described very well in the literature.[1,3] The temperature of the γ ↔ γ′ second-order transition lies between 874 °C and 779 °C. The higher temperature, which corresponds to the transition γ(γ′) ↔ γ(γ′) + ε, was established by, and this temperature was confirmed by the present DSC measurements (Table III samples 500_3, 550_4, 688_6).
The lower temperature, corresponding to the transition γ(γ′) ↔ γ(γ′) + β (Table IV, reaction 8) results from DSC studies using heating and cooling rates of 1 °C min−1 in this work. Figure 8 shows the corresponding DSC curves. The temperature of the ordering reaction γ ↔ γ′ at the phase boundary of the two-phase field is 779.6 °C.
Revised Complete Phase Diagram
By combining all of the experimental results listed in Tables II through IV it is possible to obtain a complete binary phase diagram of the Al-Cu system, which is presented in Figure 2, with details shown in Figures 4 and 7. It agrees well with the phase diagram published by Ponweiser et al. Liu et al. and Riani et al. but contains some additional clarification and improvement of areas which were not investigated in detail in the previous studies. The invariant reactions, together with the reaction temperatures and the compositions of the reacting phases are listed in Table IV. The proposed stable phase diagram should be used as a basis for further study, such as the effect of high pressures or strong magnetic fields on the Al-Cu binary phase diagram.