2021 年 62 巻 7 号 p. 1023-1029
Al–10Si–Mg alloy have been used for die casting and additive manufacturing. In this study, solution-treated Al–10Si–Mg and Al–10Si alloys (mass%) were heat-treated at several temperatures and the linear dimensional changes arising from the heat-treatments were investigated. A theoretical model for calculating the linear dimensional change in Al–Si–Mg ternary alloys was proposed, and the theoretically and the experimentally determined linear dimensional changes were compared. It was found that the linear dimensional changes of Al–10Si–Mg and Al–10Si alloys were almost the same. In both alloys, the linear dimensional changes increased with the decrease of the heat-treatment temperature. The linear dimensional changes of Al–10Si–Mg and Al–10Si alloys agreed well with the values estimated by the proposed theoretical model.
Fig. 9 Measured and calculated linear dimensional changes of solution-treated Al–10Si–Mg alloy specimens due to heat treatment.
Al–10Si–Mg alloy die castings have been used for automotive body structural components for weight saving. Usually, heat treatment under T6 conditions, i.e., a solution treatment followed by an artificial aging, is performed to improve the mechanical properties. During the heat treatment, however, the dimension of castings varies due to the precipitation of supersaturated solid solution elements and the phase transformation of metastable phases.1–6) Therefore, the amount of these dimensional changes must be estimated before setting the casting conditions.
Al–10Si–Mg alloy is also one of the most important aluminum alloys for metal additive manufacturing (AM) techniques. Although AM products are more expensive than die castings, it is desired to use AM products for low volume productions and maintenance parts that have gone out of production. However, the AM Al–10Si–Mg alloy undergoes dimensional change during heat treatment or long-term use, since it contains a high amount of supersaturated solid solution Si in α-Al phase in as-built state. The Si concentration of α-Al phase in AM Al–10Si–Mg alloy can reach 2.2 to 2.8 at% (2.3 to 2.9 mass%)7–9) due to the high cooling rate (exceeding 106 K/s).10) Thus, the dimensional change of Al–10Si–Mg alloy is also important for AM products and desired to be investigated.
The heat-treatment-induced dimensional change in an aluminum alloy can be determined both experimentally, by measuring the volumetric (density) or linear (length) change,1–6) and theoretically, by calculating the total volume change, i.e., the sum of the volume changes of all phases in the alloy, with the help of the relationship between the lattice constants of α-Al phase and the concentrations of solid solution elements.3,5,6) In our previous study,5) a theoretical model was proposed for calculating the linear dimensional change arising from heat treatments in Al–Si binary alloys. In this study, the model was extended to the Al–Si–Mg ternary system alloys. Moreover, the dimensional changes arising from heat treatments in solution-treated Al–10Si–Mg alloy specimens were also measured. Because the precipitation sequence in Al–Si–Mg alloys is complicated,11,12) heat treatments were conducted for a time long enough for the thermodynamically stable β-Mg2Si phase to precipitate completely. The experimental linear dimensional changes were compared with the values calculated by the model.
Gravity castings of Al–10Si–Mg and Al–10Si alloys were prepared from aluminum with a purity of 99.8 mass%, Al–25Si and Al–20Mg alloys. The weighed materials were melted at 1023 K and degassed in a vacuum furnace at 1023 K for 1200 s, then the melt was poured into a steel mold preheated to 623 K. The chemical compositions of the castings are listed in Table 1. The content of hydrogen in the melt measured by Ransley’s method13) was 0.20 cm3/100 g-Al. The castings were machined to small pieces (hereafter specimen) for heat treatment as shown in Fig. 1. The as-cast specimens were solution-treated at 803 K for 86.4 ks (24 h) and water-quenched. This solution-treatment time of 86.4 ks was adopted to ensure a sufficient homogenization of the specimens. The density of each solution-treated specimen was measured within 10 min after water-quenching to avoid the influence of natural aging. The solution-treated specimens were heat-treated at 473 K for 1800 ks (500 h), at 573 K for 518 ks (144 h), at 623 K for 518 ks (144 h), at 703 K for 173 ks (48 h) and at 713 K for 28.8 ks (8 h), respectively. The density of each specimen was measured at certain intervals during heat treatment. Meanwhile, some of the solution-treated specimens were naturally aged at 296 ± 1 K for 1100 ks (2 weeks), and the density change due to the natural aging was also measured.
Dimensions of gravity-castings and specimens for heat treatment.
The densities of the specimens were measured by the Archimedes method. The specimen was weighed in both air and water using a balance with an indication accuracy of 0.0001 g (H31AR, Mettler). Ion-exchanged water was used for the measurement in water. The temperature for density measurements ranged between 293 K and 296 K. The volume of the specimens at 298 K was calculated on the basis of the density of the water and the thermal expansion coefficient of the specimen (2.1 × 10−5 K−1). The density of a specimen at 298 K is obtained by dividing its weight in air by its volume at 298 K.
The linear dimensional change ΔL [%] induced by heat treatment or natural aging is given by eq. (1).
| \begin{equation} \Delta L = \left[\left(\frac{\rho^{\text{A}}}{\rho^{\text{B}}} \right)^{1/3} {}- 1 \right] \times 100 \end{equation} | (1) |
The microstructures and the Si and Mg distributions in the Al–10Si–Mg alloy specimens were investigated by electron probe microanalysis (EPMA). Scanning transmission electron microscopy, STEM (JEM-2100F, JEOL), equipped with energy dispersive X-ray spectroscopy (EDS) was employed to identify the precipitates in the Al–10Si–Mg alloy specimens. An ion slicer (EM-09100IS, JEOL) was used to fabricate the samples for TEM analysis.
The dimensional changes resulting from the precipitation of solid solution elements during heat treatment were calculated theoretically for solution-treated specimens. Denoting the state of the specimen before and after heat treatment by A and B, respectively, the linear dimensional change is given by eq. (2).
| \begin{equation} \Delta L = \left[\left(\frac{V^{\text{B}}}{V^{\text{A}}} \right)^{1/3} {}- 1 \right] \times 100 \end{equation} | (2) |
In equilibrium state, Al–Si–Mg alloys are composed of α-Al, Si (which has a diamond crystal structure), and β-Mg2Si phases. Thus,
| \begin{equation} V^{\text{A}} = V_{\alpha}^{\text{A}} + V_{\text{dia}}^{\text{A}} + V_{\beta}^{\text{A}} \end{equation} | (3) |
| \begin{equation} V_{\alpha}^{\text{A}} = P_{\alpha} \times (m_{\text{${\alpha}$(Al)}}^{\text{A}} + m_{\text{${\alpha}$(Si)}}^{\text{A}} + m_{\text{${\alpha}$(Mg)}}^{\text{A}}) \end{equation} | (4a) |
| \begin{equation} V_{\text{dia}}^{\text{A}} = P_{\text{dia}} \times m_{\text{dia(Si)}}^{\text{A}} \end{equation} | (4b) |
| \begin{equation} V_{\beta}^{\text{A}} = P_{\beta} \times (m_{\text{${\beta}$(Si)}}^{\text{A}} + m_{\text{${\beta}$(Mg)}}^{\text{A}})/3 \end{equation} | (4c) |
| \begin{equation} P_{\alpha} = (d_{\alpha}^{\text{A}})^{3} \times \mathrm{N}/n_{\alpha} \end{equation} | (5a) |
| \begin{equation} P_{\text{dia}} = (d_{\text{dia}}^{\text{A}})^{3} \times \mathrm{N}/n_{\text{dia}} \end{equation} | (5b) |
| \begin{equation} P_{\beta} = (d_{\beta}^{\text{A}})^{3} \times 3 \times \mathrm{N}/n_{\beta} \end{equation} | (5c) |
| \begin{equation} d_{\alpha}^{\text{A}} = d_{\alpha}^{0} + \frac{\partial d_{\alpha}}{\partial x_{\text{Si}}}x_{\text{Si}}^{\text{A}} + \frac{\partial d_{\alpha}}{\partial x_{\text{Mg}}}x_{\text{Mg}}^{\text{A}} \end{equation} | (6) |
The molar amounts are given by
| \begin{equation} m_{\text{${\alpha}$(Al)}}^{\text{A}} = \frac{(1 - C_{\text{${\alpha}$(Si)}}^{\text{A}} - C_{\text{${\alpha}$(Mg)}}^{\text{A}})W_{\alpha}^{\text{A}}}{\mathrm{M}_{\text{Al}}} \end{equation} | (7a) |
| \begin{equation} m_{\text{${\alpha}$(Si)}}^{\text{A}} = \frac{C_{\text{${\alpha}$(Si)}}^{\text{A}}W_{\alpha}^{\text{A}}}{\mathrm{M}_{\text{Si}}} \end{equation} | (7b) |
| \begin{equation} m_{\text{${\alpha}$(Mg)}}^{\text{A}} = \frac{C_{\text{${\alpha}$(Mg)}}^{\text{A}}W_{\alpha}^{\text{A}}}{\mathrm{M}_{\text{Mg}}} \end{equation} | (7c) |
| \begin{equation} m_{\text{dia(Si)}}^{\text{A}} = \frac{W_{\text{dia}}^{\text{A}}}{\mathrm{M}_{\text{Si}}} \end{equation} | (7d) |
| \begin{equation} m_{\text{${\beta}$(Si)}}^{\text{A}} = \frac{C_{\text{${\beta}$(Si)}}^{\text{A}}W_{\beta}^{\text{A}}}{\mathrm{M}_{\text{Si}}} \end{equation} | (7e) |
| \begin{equation} m_{\text{${\beta}$(Mg)}}^{\text{A}} = \frac{C_{\text{${\beta}$(Mg)}}^{\text{A}}W_{\beta}^{\text{A}}}{\mathrm{M}_{\text{Mg}}} \end{equation} | (7f) |
Si and Mg solubilities in α-Al phase of Al–10Si–Mg (Al–9.8Si–0.39Mg) alloy calculated by Thermo-calc software with the TCAL4 database.
It is well known that the lattice constant of a solid solution phase in a binary alloy is approximately proportional to the atomic fraction of solid solution elements. The lattice constants, dα, of homogeneous α-Al phases containing solid solution Si or Mg obtained by quenching from high temperature have been reported up to 0.93 at% Si for Al–Si alloys and 7.63 at% Mg for Al–Mg alloys.15) By fitting the data to eq. (8),
| \begin{equation} d_{\alpha} = d_{\alpha}^{0} + x_{\text{X}}\left(\frac{\partial d_{\alpha}}{\partial x_{\text{X}}} \right) \end{equation} | (8) |
Lattice constants of α-Al phase in Al–Si–Mg ternary alloys determined by XRD measurement15) and estimated in this study for concentrations of up to 1.42 at% Si and 1.42 at% Mg. Lattice constant of pure Al was taken to be 0.40495 nm.
The solid solubility extension achieved by rapid solidification can be determined from the lattice constant of α-Al phase.21) Al–Si alloys, however, exhibit a large variation in the relationship between the lattice constant and the concentration of solid solution Si at high supersaturation levels, as shown in Fig. 4. A problem to note in measuring the concentration of supersaturated solid solution Si is the formation of very fine Si precipitates.17,21) To solve this problem, Shingu et al. applied a splat cooling technique to fabricating rapidly solidified Al–Si alloy specimens with homogeneous α-Al phase.20) The absence of precipitated Si in the specimens was confirmed by TEM up to 16 at% Si in the alloy. The lattice constants of these specimens are plotted in Fig. 4. Clearly, the relationship used in the present study agrees with the experimental data obtained by Shingu et al.
Lattice constants of α-Al phase in Al–Si binary alloys for high concentrations of solid solution Si. The lattice constants reported by Shingu20) were determined from rapidly solidified specimens free from Si precipitates. The relationship used in the present study agreed with the experimental results by Shingu.
The SEM and EPMA images of the as-cast and solution-treated Al–10Si–Mg alloy specimens are shown in Fig. 5. Segregation of Si and Mg was observed in the as-cast specimen. The concentrated Mg in the inter-dendrite region is presumed to be β-Mg2Si phase having formed during the eutectic solidification of α-Al, Si, and β-Mg2Si phases. This microscopic segregation was completely homogenized by the solution treatment.
Si and Mg distributions in Al–10Si–Mg alloy specimens in as-cast and solution-treated states.
The density of the Al–10Si–Mg alloy specimen heat-treated at 473 K for 1800 ks is listed in Table 4. The density was 2661 kg/m3 in as-cast state and increased to 2663 kg/m3 after the solution-treatment (ST) due to the increase of solid solution Si in α-Al phase. In contrast, the density decreased to 2655 kg/m3 after the heat treatment due to the precipitation of supersaturated solid solution elements.
The linear dimensional change due to the natural aging is shown in Fig. 6. Although the solution-treated Al–10Si–Mg alloy specimens showed a gradual growth, their linear dimensional changes were quite small. This means that the linear dimensional change caused by natural aging before the first density measurement was negligible. Figure 7 shows the linear dimensional change arising from the heat treatments at 473 K and 713 K. The Al–10Si–Mg alloy specimens grew more rapidly than the Al–10Si alloy specimens in the early stage of the heat treatment at 473 K. However, the amounts of the linear dimensional change in these specimens became almost the same by the end of the heat treatments. The final linear dimensional changes by the heat treatments at 473 K and 713 K were 0.10% and 0.07% for the Al–10Si–Mg alloy specimens, and 0.10% and 0.06% for the Al–10Si alloy specimens, respectively.
Linear dimensional changes of solution-treated Al–10Si–Mg and Al–10Si alloy specimens due to natural aging at 296 K.
Linear dimensional changes of solution-treated Al–10Si–Mg and Al–10Si alloy specimens due to heat treatment at (a) 473 K and (b) 713 K.
Figure 8 shows the microstructure of the Al–10Si–Mg alloy specimen heat-treated at 473 K for 1800 ks, obtained by STEM. The microstructures denoted as 1 and 2 in the STEM-bright field (BF) image (Fig. 8(a)) are α-Al matrix. The concentrations of Si and Mg in α-Al matrix are very low, as shown in Figs. 8(b) and 8(c), which agree with the solubility limits of these elements in Al–10Si–Mg alloy at 473 K given in Fig. 2. The precipitates labelled as 3 and 4 in Fig. 8(a) are β-Mg2Si phase. The lattice constants of precipitate 3 determined from the selected area electron diffraction (SAED) pattern are 0.23 nm and 0.29 nm (OA and OB in Fig. 8(d)), which are comparable to the distances between lattice planes of the cubic structure of β-Mg2Si phase, d220 = 0.2245 nm and d200 = 0.3176 nm, respectively. The SAED pattern of precipitate 4 (Fig. 8(e)) corresponds to that of the β-Mg2Si precipitate reported by Kanno,22) the crystallographic relation of which is (100)Mg2Si // (100)α-Al and [100]Mg2Si // [001]α-Al. The chemical analyses of precipitates 3 and 4 (Fig. 8(c)) indicate that the atomic ratio of Mg to Si is approximately 2:1, although the data contain signals coming from both β-Mg2Si phase and α-Al matrix. Precipitates 5 and 6 are considered to be Si phase, given that no Mg signal is detected. These results indicate that the Al–10Si–Mg alloy specimen heat-treated at 473 K for 1800 ks is composed of the thermodynamically stable phases, α-Al, Si, and β-Mg2Si.
Microstructure of Al–10Si–Mg alloy specimen heat-treated at 473 K for 1800 ks. (a) STEM bright-field image, (b) STEM dark-field image and EDS mappings of the same area, (c) EDS point analyses, (d), (e) SAED patterns of precipitates 3 and 4 indicated in (a).
The measured final linear dimensional changes of the heat-treated specimens are shown in Fig. 9 together with that calculated by the theoretical model. The linear dimensional change increases with the decrease of heat treatment temperature, but the linear dimensional change below 573 K is almost constant. It is also found that the difference of the linear dimensional changes between the Al–10Si–Mg alloy and the Al–10Si alloy is small. The theoretically estimated dimensional changes agree well with the experimentally measured values for both Al–10Si–Mg and Al–10Si alloys, indicating that the theoretical model is appropriate for predicting the linear dimensional change of Al–Si–Mg ternary system alloys.
Measured and calculated linear dimensional changes of solution-treated Al–10Si–Mg alloy specimens due to heat treatment.
The linear dimensional changes of the solution-treated Al–10Si–Mg and Al–10Si alloy specimens mainly depend on the Si solubility in α-Al phase. Thus, the difference between the theoretically estimated linear dimensional changes of the Al–10Si–Mg and Al–10Si alloy specimens is small (Fig. 9). However, the increase in the linear dimensional change of the Al–10Si–Mg alloy specimens was faster than that of the Al–10Si alloy specimens at the early stage of the heat treatment at 473 K (Fig. 7(a)). In Al–Si alloys, the supersaturated solid solution Si can precipitate both in α-Al phase and on the surface of eutectic Si. Dong et al.6) observed that the addition of Mg into Al–11Si alloys increased the number of Si precipitates in α-Al phase during heat treatment at 473 K and accelerated the dimensional change of the alloys. In the Al–10Si–Mg alloy specimen, Mg is presumed to have promoted the precipitation of supersaturated solid solution Si in α-Al phase.
5.2 Effect of Fe on linear dimensional changeThe Al–10Si–Mg alloy specimens contain 0.11 mass% Fe as listed in Table 1. In these specimens, β-AlFeSi phases may precipitate during solidification and transform to α-AlFeSi phases during the solution treatment. Thus, the solution-treated Al–10Si–Mg alloy specimens used in this study may contain α-AlFeSi phases. For estimating the effect of Fe on the linear dimensional change, the phase constitution of Al–10Si–Mg alloy (Al–9.8Si–0.39Mg–0.11Fe) at equilibrium state was determined using the Thermo-calc version S with the TCAL4 database as shown in Fig. 10, supposing the α-AlFeSi phase to be in the form of, e.g., α-Al8Fe2Si. The content of α-Al8Fe2Si phase in the solution-treated (803 K) and the heat-treated (473 K) Al–10Si–Mg alloy specimens are almost the same and less than 0.4 mass%. Therefore, the experimental error due to the Fe impurity is very small.
Equilibrium phase constitution in Al–10Si–Mg (Al–9.8Si–0.39Mg–0.11Fe) alloy calculated with the Thermo-calc software.
The Si concentration of α-Al phase in the as-built state AM Al–10Si–Mg alloy can reach 2.2 to 2.8 at% (2.3 to 2.9 mass%).7–9) By the theoretical model, the dimensional change of Al–10Si–Mg alloy containing such a high level of supersaturated solid solution Si can also be calculated. When the Si concentration of α-Al phase is 2.3 mass%, the maximum dimensional change due to heat treatment below 673 K is expected to be 2.2%. This indicates that the dimensional change of the as-built state AM Al–10Si–Mg alloy is much higher than that of the solution-treated Al–10Si–Mg alloy. Therefore, a detailed investigation of the dimensional change of AM Al–10Si–Mg is desired.
The solution-treated Al–10Si–Mg and Al–10Si alloy specimens were heat-treated at several temperatures and the resulting linear dimensional changes were investigated. Moreover, a theoretical model for estimating the linear dimensional change in the Al–Si–Mg ternary alloy was developed and the calculated linear dimensional changes were compared with the experimentally measured values.
