Dissolution Mechanism of Intermetallic Layer by Iron Erosion in Aluminum-Based Molten Binary Alloys

Abstract

Erosion testing was conducted on iron specimens in various aluminum-based molten binary alloys, and the correspondence between the intermetallic compounds formed at the contact interface and the thermodynamically stable phases was examined. On the basis of the results, the mechanism and dominant factor of the dissolution of the intermetallic layers were investigated. Although the erosion ratio of the specimens by molten Al–3%Mn alloy was significantly small, the ratios by molten alloys of the Al–Si system were great, in comparison to those by other melts. Meanwhile, the intermetallic compounds identified by EBSD corresponded to the stable phases based on the equilibrium calculation of the compositions analyzed by EDS and/or the pseudo binary phase diagrams of the aluminum alloy-Fe systems. In addition, there was a positive correlation between the experimental erosion rate and apparent saturation solubility of Fe to the melts based on the pseudo binary phase diagrams. These results suggest that the dissolution of the layers was caused by almost the same mechanism as the phenomena described by Noyes-Whitney-Nernst equation, and that the saturation solubility of Fe is a dominant factor affecting the diffusion-controlled dissolution from the solid-liquid interfaces under local equilibrium.

 

This Paper was Originally Published in Japanese in J. JFS 93 (2021) 541–550.

1. Introduction

When ferrous materials and molten aluminum alloys form an interface, wastage and shape variation in ferrous components often occur, which is known as erosion. For example, the erosion of alloy tool steel members such as molds, core pins, plunger tips, and shot sleeves causes issues in die casting processes. Surface modification of ferrous materials such as nitriding and oxidation treatments have been widely investigated to improve their erosion resistance,^1–6) and some of these techniques have been industrially utilized. Ferrous composites reinforced by excellent erosion-resistant materials such as boride and cermet have also been investigated.^7–9) Meanwhile, the erosion resistance of materials is evaluated by erosion testing in a molten aluminum alloy. The ferrous materials employed as the specimens are usually multicomponent alloys.^10–14) In addition, erosion testing is conducted under various conditions, such as static insuccation and rotation in the melt.^1–14) Multicomponent applicative aluminum alloys are also often employed as the melt.^3–12,14) These studies suggest that complex effects from various factors contribute to erosion testing and that the systematic determination of erosion resistance and its practical utilization based on the results of erosion tests has been a challenge.

In general, the formation of intermetallic layers at the interface between molten aluminum alloys and ferrous materials and the dissolution and/or delamination of these intermetallic layers progress the erosion.^11,15–17) Thus, the formation behavior of the intermetallic layers is one of the dominant factors affecting the erosion resistance of ferrous materials.^12,15–18) For the contact interface between pure molten aluminum and commercially pure iron or low-carbon steel, stable phases in the Al–Fe binary phase diagram are formed.^13,17–21) Meanwhile, for the contact interface between the molten aluminum alloy and ferrous material, the calculated pseudo-binary phase diagram of the aluminum alloy–ferrous material system has the potential to reflect the generation behavior of the intermetallic compounds. In addition, the relation not only between the pseudo-binary phase diagram and the morphology of the intermetallic layers but also between the diagrams and the diffusion or dissolution behavior of the layers is suggested. The clarification of these detailed mechanisms can enable us to quantitatively characterize the erosion behavior associated with the dissolution of the intermetallic layers using pseudo-binary phase diagrams and to utilize thermodynamic calculations for the guideline of the development of erosion-resistant materials. In this study, we conducted the erosion testing of iron specimens in various Al-based molten binary alloys and examined the links between the constituent phases of the intermetallic layers and the thermodynamically stable phases or pseudo-binary phase diagrams. Based on the obtained results, we investigated the factors and the mechanism affecting the dissolution behavior of the intermetallic layers.

2. Experimental and Numerical Methods

2.1 Erosion testing

Figure 1 shows the schematic of erosion testing. This testing was conducted by rotating three iron specimens in molten aluminum, as shown in Fig. 1. The upside of the specimens (except for the testing portions) was covered with alumina cement to keep the contact length between the specimens (testing portions) and the melt constant at 50 mm. The rotation speed was set to 200 rpm (rotation radius of 18 mm), and the weight of the melt was 1800 g, irrespective of alloy compositions. JIS R 2701 No. 6 graphite crucible (capacity of 860 ml) was used. Before the start of erosion testing, the specimens were held near the melt for approximately 5 min to avoid rapid temperature change in the melt. After testing, aluminum adhered to the specimens was removed by resolving (immersing until the generation of hydrogen bubbles almost ended) in 20% sodium hydroxide aqueous solution (approximately 50°C). By weighing the specimens before and after testing, the erosion ratio (weight change ratio of the testing portion) was calculated using eq. (1):   

\begin{equation} E\ (\%) = \frac{W_{\text{b}} - W_{\text{a}}}{W_{\text{b}}} \times 100 \end{equation} (1)

Here, E is the erosion ratio, and W_b and W_a are the weights of the tested portion before and after testing, respectively.

Fig. 1

Schematic of erosion testing.

Table 1 shows the composition of molten aluminum used in erosion testing.²²⁾ Pure aluminum (99.99% purity) and Al-based binary alloys prepared by mixing pure aluminum and other pure metals or Al-based binary master alloys were used as melts. For Al–Cr, Al–Co, and Al–Mn alloys, Cr, Co, and Mn contents were determined based on the corresponding calculated binary phase diagrams to avoid the generation of solid phases under the testing temperature. Bars (diameter: 10 mm, surface: glossy) of commercial low-carbon steel (The Nilaco Corporation, chemical composition: Fe–0.18% C–0.05% Si–0.59% Mn–0.012% P–0.041% S) or carbon steel for machine structural use JIS S20C (JFE Bars & Shapes Corporation, chemical composition: Fe–0.18% C–0.19% Si–0.40% Mn–0.014% P–0.016% S–0.12% Cu–0.05% Ni–0.07% Cr, as hot-rolled) were cut and used as specimens. Although these two different steels were separately employed for testing, obtained results were collectively used as the results of iron specimens (approximately 99% purity) because the differences between their chemical compositions and Fe contents were minor, and the erosion ratio results were almost the same. The results of the specimens with and without cleaning the testing portions by acetone after covering with the alumina cement were also collectively used because no significant differences were obtained in the erosion ratio results. The testing temperature was set to 700°C, and the testing time was mainly set to 2 h. For the pure molten aluminum, erosion testing was also conducted for 1 and 3 h.

Table 1 Molten aluminum used in erosion testing.

2.2 Observations and analyses

The polished cross sections of the specimens without the removal of aluminum adhered by erosion testing were observed using a scanning electron microscope (SEM). The quantitative analysis of phase compositions and the mapping of elemental distributions were conducted using energy-dispersive X-ray spectroscopy (EDS) and an electron probe microanalyzer (EPMA), respectively. Through the separate observations and analyses of the specimens after the removal of the adhered aluminum, the removal of most intermetallic layers by dissolving in a sodium hydroxide aqueous solution was confirmed, irrespective of the melt compositions.¹¹⁾

The cross sections of the specimens without the removal of aluminum adhered by erosion testing were also polished using a vibratory polisher and colloidal silica suspension. For these polished cross sections, the crystal structures of constituent phases in the vicinity of the contact interfaces between the molten aluminum and the specimens were analyzed using electron backscatter diffraction (EBSD). The phase mapping based on EBSD analysis was also conducted. Crystal structure analysis enables the determination of plausible phases by comparing the EBSD patterns with the data of various crystal systems. In this study, the identification of constituent phases was performed as follows. First, analysis using the all available data of crystal systems was conducted, and undetected phases were excluded from candidate phases. Second, the candidate phase with the maximum ranking factor calculated from the vote number based on the voting method, fit index, and reliability index was identified as the most likely phase.²³⁾ Although this procedure makes it difficult to discriminate the lattice parameters and phases with the same crystal structures, the constituent phases analyzed in this study were identified according to the candidate phases. Here, the Si phase that may have been formed during and after the solidification of molten Al–Si binary alloys was excluded from the candidate phases because it was difficult to be discriminated from the Al phase and α-Fe phase. For crystal structure and phase analyses, the step size was adjusted based on magnification, and data processing by a clean-up function was not conducted.

2.3 Thermodynamic calculations

Equilibrium calculations of binary and ternary systems and phase diagram calculations of binary and pseudo-binary systems were performed using a commercial thermodynamic calculation software (Thermo-Calc Software AB, Thermo-Calc) and an Al-based alloy database (Thermo-Calc Software AB, TCAL4) based on the CALPHAD method.²⁴⁾ The pseudo-binary phase diagram was constructed for an Al-based binary alloy (Al–y% X, y: X content) and Fe while decreasing in Al and X with an increase in Fe and maintaining a constant ratio between Al and X (Al:X = (100 − y):y). This is referred to as the pseudo-binary phase diagram of the (Al–y% X)–Fe system. For these calculations except where specifically noted, thermodynamically stable phases were determined from the calculation of all phases available in the database. The apparent saturation solubility of Fe in the melts under the testing temperature (maximum Fe content that maintains the single stable liquid phase in the pseudo-binary phase diagram) was obtained from the pseudo-binary phase diagram. The calculation results using volume data included in the thermodynamic database were also utilized as needed.

3. Results and Discussion

3.1 Erosion ratio

Figure 2 shows the erosion ratios of iron specimens under a testing time of 2 h. The erosion ratios in molten Al–3.25% Mg, Al–6.8% Zn, Al–0.5% Cr, and Al–4.5% Cu alloys were almost the same as that in pure molten aluminum, as shown in Fig. 2. In comparison with these erosion ratios, the erosion ratio in molten Al–1% Co alloy was smaller, and the erosion ratio in molten Al–3% Mn alloy was significantly smaller. On the other hand, the erosion ratio in molten Al–Si alloys was greater than that in pure molten aluminum or other Al-based molten alloys. Although the magnitude of the erosion ratios in molten Al–3–10.8% Si alloys corresponds to the Si content, the erosion ratio in molten Al–17% Si alloy was smaller than those in molten Al–7% Si and Al–10.8% Si alloys.

Fig. 2

Erosion ratio of iron specimens under testing time of 2 h.

Figure 3 shows the relationships between the erosion ratio or the erosion thickness of the iron specimens in pure molten aluminum and testing time. Here, the erosion thickness was calculated according to the erosion ratio under the assumption that erosion is caused via a uniform reduction in dimensions along the normal directions of both side and bottom surfaces (the testing portion maintains the circular cylindrical shape). As shown in Fig. 3, the erosion ratio and the erosion thickness changed with the testing time that can be fitted by power approximation (least-squares approximation of double logarithmic plot). In particular, the exponent of the power approximation of the relationship between the erosion ratio and the testing time was 0.4933 that was very close to 0.5. Figure 4 shows the cross-sectional SEM images of these specimens after testing. Intermetallic layers were observed at the interface between the solidified melt and the specimen, as shown in Fig. 4. Although the thickness of the intermetallic layers was gradually increased during testing, the delamination of the layers was not observed. These results suggest that the erosion thickness shown in Fig. 3 is the sum of the dissolved thickness in the melt and the thickness of the formed intermetallic layers, and that the erosion thickness represents the average value because the actual thicknesses are not fully uniform in the testing portions because of effects such as the flow of the melt.¹¹⁾ Figure 4 shows that the thickness of the intermetallic layers was approximately 50–200 µm under the testing time of 1–3 h and was smaller than the dissolved thickness. The margin between the erosion thickness and the thickness of the intermetallic layers similarly exhibited the changes in time dependence that can be fitted by power approximation. These results suggest the possibility that the process obeying the parabolic law (being proportional to the square root of time) is significantly related to the erosion behavior, and that the diffusion process in the melt or the intermetallic layers may be the rate-controlling step of erosion.²⁵⁾

Fig. 3

Relationships between erosion ratio or erosion thickness of specimens and testing time.

Fig. 4

Cross-sectional SEM images in vicinity of interface between pure molten Al and specimen under testing time of (a) 1 h, (b) 2 h and (c) 3 h ((a) and (c): SE images, (b): BSE image).

3.2 Constituent phases of intermetallic layers

Figures 5 and 6 show the examples of the cross-sectional backscattered electron (BSE) images of the specimens after erosion testing in the Al-based molten binary alloys and the element distribution by EPMA in the vicinity of the contact interfaces, respectively. For molten Al–4.5% Cu, Al–3.25% Mg, and Al–6.8% Zn alloys, intermetallic layers with rounded convex portions, which significantly protruded to the specimen side, were observed, as shown in Figs. 5(a), 5(b), and 5(c), respectively. The morphology of these layers was almost the same as that for molten pure aluminum shown in Fig. 4(b). For molten Al–0.5% Cr alloy, the morphology of the intermetallic layers was almost the same as that for pure molten aluminum or molten Al–4.5% Cu, Al–3.25% Mg, and Al–6.8% Zn alloys, and the convex portions with high Cr content protruding to the melt side were also scattered at the solid–liquid interface, as shown in Fig. 6(a). For molten Al–1% Co and Al–3% Mn alloys, the convex portions with high Co and Mn contents protruding to the melt side were markedly formed, as shown in Figs. 6(b) and 6(c), respectively. In particular, for the molten Al–3% Mn alloy, a phase with high Mn content entirely covered the solid–liquid interface. On the other hand, for the molten Al–7–17% Si alloys, two thin intermetallic layers were observed, while the convex portions protruding to the specimen were not observed, as shown in Figs. 5(d), 5(e), and 5(f). For the molten Al–3% Si alloy, two intermetallic layers that were similar to those in the molten Al–7–17% Si alloys were observed, and the differences in the contents of the elements were not clear, as shown in Fig. 6(d). The delamination of the intermetallic layers was not observed, irrespective of the melt compositions.

Fig. 5

Cross-sectional BSE images in vicinity of interface between molten (a) Al–4.5% Cu, (b) Al–3.25% Mg, (c) Al–6.8% Zn, (d) Al–7% Si, (e) Al–10.8% Si, (f) Al–17% Si alloys and specimen under testing time of 2 h.

Fig. 6

Cross-sectional element distribution in vicinity of interface between solidified molten (a) Al–0.5% Cr, (b) Al–1% Co, (c) Al–3% Mn, (d) Al–3% Si alloys and specimen under testing time of 2 h.

Figure 7 shows the examples of phase distributions (phase maps) based on EBSD analysis in the vicinity of the contact interfaces. Table 2 shows the principal constituent phases of the intermetallic layers based on EBSD analysis and the chemical compositions analyzed by EDS. Here, the step size of the EBSD analysis shown in Figs. 7(a)–7(f) was set to 0.5, 1, 0.3, 0.5, 0.5, and 0.3 µm, respectively. The expediential division and number of the layers are also shown in Fig. 7 and Table 2. For pure molten aluminum shown in Table 2, the layer located near the presumable solid–liquid interface during testing (melt side) and the neighboring (specimen side) tongue-like convex layer was identified as Al₁₃Fe₄ (also known as Al₃Fe) and Al₅Fe₂, irrespective of the testing time, which have been reported previously.^13,17–21) For the molten Al–0.5% Cr alloy, the constituent phases of the layers were also the same as those of pure molten aluminum. Although the constituent phases of the molten Al–3.25% Mg, Al–6.8% Zn, and Al–4.5% Cu alloys have not been identified by EBSD, the morphology of the intermetallic layers shown in Figs. 5(a)–5(c) and the compositions of these layers analyzed by EDS were almost the same as those of pure molten aluminum and the molten Al–0.5% Cr alloy shown in Figs. 4(b) and 6(a), respectively. These results suggest that the constituent phases of the intermetallic layers for the molten Al–3.25% Mg, Al–6.8% Zn, and Al–4.5% Cu alloys are Al₁₃Fe₄ and Al₅Fe₂. For the molten Al–1% Co alloy shown in Fig. 7(a), the constituent phases were almost the same, and the phase of the convex portions protruding to the melt was identified as Al₉Co₂ via EBSD and EDS analyses. For the molten Al–3% Mn alloy shown in Fig. 7(b), the constituent phases of the main two layers were also the same as the two phases observed for pure molten aluminum and other molten alloys (Al₁₃Fe₄ and Al₅Fe₂), and the phase of the melt-side layer was identified as Al₆Mn. On the other hand, for the phase distribution of the molten Al–Si alloys, not only Al₁₃Fe₄ and/or Al₅Fe₂ but also Al₈Fe₂Si (also known as α-AlFeSi, τ₅,^26,27) and Al_7.4Fe₂Si^26,28)) and/or Al₉Fe₂Si₂ were observed, as shown in Figs. 7(c), 7(d), 7(e) and 7(f). The atomic ratios of the intermetallic compounds mentioned above and below are not always the same as their nominal ratios, because the compounds often allow a range of mass percentages of contents of the elements such as Fe and Si.^29–31)

Fig. 7

Cross-sectional phase distribution (phase map) based on EBSD analysis in vicinity of interface between solidified molten (a) Al–1% Co, (b) Al–3% Mn, (c) Al–3% Si, (d) Al–7% Si, (e) Al–10.8% Si, (f) Al–17% Si alloys and specimen under testing time of 2 h.

Table 2 Relationship between principal intermetallic compounds based on EBSD analysis in vicinity of interfacial layers under testing time of 2 h and principal stable phases based on thermodynamic calculation under testing temperature.

3.3 Correspondence with thermodynamically stable phases

Figure 8 shows the aluminum-rich portion of the binary phase diagram of the Al–Fe system and the Al alloy-rich portions of the pseudo-binary phase diagrams of the Al-based binary alloy–Fe systems. Table 2 also shows the principal stable and metastable phases of the intermetallic layers based on the thermodynamic calculations of the analyzed chemical compositions and the principal stable phases in the pseudo-binary phase diagrams under the testing temperature. Here, the stable phases under a volume fraction of more than 2 vol% and under the composition range less than the Fe content that leads to the generation of Al₅Fe₂ in the Al-rich or Al alloy-rich portion of the pseudo-binary phase diagrams are shown in Table 2. The analyzed contents shown in Table 2 are the averages of several analytical values. As shown in Table 2, the above-mentioned constituent phases of the intermetallic layers mostly corresponded to the stable phases determined based on the equilibrium calculations of the analyzed compositions and pseudo-binary phase diagrams. Since the assessed ternary data of Al–Fe–Cr and Al–Fe–Co systems are not included in the thermodynamic database, the solid solutions of Cr or Co in Al₁₃Fe₄ or Al₅Fe₂ and Fe in Al₉Co₂, as well as the ternary interaction parameters are disregarded in the thermodynamic calculations. The results shown in Table 2 suggest that the accuracy of the thermodynamic calculations for the molten Al–0.5% Cr and Al–1% Co alloys was practically sufficient. Although the main phase of the melt-side intermetallic layer based on the analyzed composition for the molten Al–10.8% Si alloy was τ₂ (also known as γ-AlFeSi, Al₃FeSi,^26,28,31) and Al₅Fe₂Si₂²⁷⁾), the morphology was almost the same as that of the molten Al–7% Si alloy based on Figs. 5(d), 5(e), 7(d), and 7(e). Not only these results but also the metastable phases based on the equilibrium calculation shown in Table 2 and the pseudo-binary phase diagram of the (Al–10.8% Si)–Fe system shown in Fig. 8(j) suggest that the main phase of the melt-side intermetallic layer for the molten Al–10.8% Si alloy is Al₈Fe₂Si. In addition, the convex Al₉Fe₂Si₂ protruding to the melt in the phase distribution for the molten Al–7% Si and Al–10.8% Si alloys could be generated during the cooling process after testing¹¹⁾ because this phase was the main stable phase based on the equilibrium calculations of the analyzed compositions under approximately 600°C. Meanwhile, as shown in Figs. 5(f) and 7(f), the morphology of the melt-side intermetallic layer for the molten Al–17% Si alloy was different from that of the molten Al–7% Si and Al–10.8% Si alloys shown in Figs. 5(d), 5(e), 7(d), and 7(e). The analyzed composition of the phase of the melt-side intermetallic layer for the molten Al–17% Si alloy was also close to the composition of τ₄ (also known as δ-AlFeSi, Al₃FeSi₂,^19,26–28) and Al₃FeSi₃²⁶⁾). These results imply that the main phase of the melt-side intermetallic layer for the molten Al–17% Si alloy is τ₄ and that the phase was difficult to be identified by EBSD analysis. However, the range of Fe content under the testing temperature where τ₄ is stable in the pseudo-binary phase diagram of the (Al–17% Si)–Fe system shown in Fig. 8(k) was narrow, and the volume fraction of τ₄ was less than approximately 1.38 vol%. Furthermore, the stable phases based on the equilibrium calculations of the analyzed compositions sometimes included a liquid phase. This suggests that the intermetallic compounds and liquid coexisted in the analyzed positions during testing.

Fig. 8

(a) Al-rich portion of binary phase diagram of Al–Fe system and Al alloy-rich portions of pseudo binary phase diagrams of (b) (Al–3.25% Mg)–Fe, (c) (Al–6.8% Zn)–Fe, (d) (Al–0.5% Cr)–Fe, (e) (Al–4.5% Cu)–Fe, (f) (Al–1% Co)–Fe, (g) (Al–3% Mn)–Fe, (h) (Al–3% Si)–Fe, (i) (Al–7% Si)–Fe, (j) (Al–10.8% Si)–Fe and (k) (Al–17% Si)–Fe systems.

The Al alloy-rich portions of the pseudo-binary phase diagrams of the (Al–3.25% Mg)–Fe and (Al–6.8% Zn)–Fe systems shown in Figs. 8(b) and 8(c), respectively, were similar to the Al-rich portion of the Al–Fe binary phase diagram shown in Fig. 8(a). Although the Al alloy-rich portions of the pseudo-binary phase diagrams of the (Al–0.5% Cr)–Fe and (Al–4.5% Cu)–Fe systems shown in Figs. 8(d) and 8(e), respectively, included three-phase areas consisting of a liquid phase and intermetallic compounds, such as Al₄₅Cr₇, Al₈Cr₅ (term in TCAL4: GAMMA_D810),³²⁾ and Al₆₂Cu₂₅Fe₁₃, the volume fraction of these compounds under the testing temperature was less than 5 vol%. These results suggest that the pseudo-binary phase diagrams of the (Al–0.5% Cr)–Fe and (Al–4.5% Cu)–Fe systems are actually almost the same as the Al–Fe binary phase diagram. The Al alloy-rich portions of the pseudo-binary phase diagrams of the (Al–1% Co)–Fe and (Al–3% Mn)–Fe systems shown in Figs. 8(f) and 8(g), respectively, also included three-phase areas involving a liquid phase and intermetallic compounds, such as Al₉Co₂ and Al₆Mn, and the volume fraction of Al₆Mn under the testing temperature was less than 5 vol%. Meanwhile, for the Al alloy-rich portion of the pseudo-binary phase diagram of the (Al–3% Mn)–Fe system, the apparent saturation solubility of Fe in the liquid phase under the testing temperature was also smaller than those in the other systems because of the generation of Al₆Mn under a range of low Fe content. Further, the Al alloy-rich portions of the pseudo-binary phase diagrams of the (Al–3% Si)–Fe, (Al–7% Si)–Fe, (Al–10.8% Si)–Fe, and (Al–17% Si)–Fe systems shown in Figs. 8(h), 8(i), 8(j), and 8(k), respectively, included three-phase areas comprising a liquid phase and intermetallic compounds, such as Al₈Fe₂Si, τ₂, τ₁, τ₁₁, and τ₄. For these pseudo-binary phase diagrams of the Al–Si binary alloy–Fe systems, the apparent saturation solubility of Fe in the liquid phase under the testing temperature was greater than those found in the other systems.

3.4 Factors affecting dissolution behavior

The above-mentioned findings indicate that the diffusion process in the melt or the intermetallic compounds is the rate-controlling step of the erosion behavior in this study.²⁵⁾ For the resolution of solids such as medicines into the water, it is known that diffusion into the aqueous solution is the rate-controlling step, and the resolution rate in this situation is described by the Noyes–Whitney–Nernst equation (eq. (2)).³³⁾   

\begin{equation} \frac{dC}{dt} = \frac{DA}{\delta V}(C_{0} - C) \end{equation} (2)

Here, dC/dt is the resolution rate, D is the diffusion coefficient, A is the surface area of solute, δ is the thickness of the diffusion layer, and V is the volume of the solvent. C₀ and C are the saturated solubility and concentration of solute in the solvent, respectively. The time variation of the thickness of the diffusion layer during the erosion and the corresponding gradient of Fe content under a quiescent condition is significant.³⁴⁾ On the other hand, the time variation of them under the condition of sufficient melt volume and agitation should be small, and this leads to a small variation in the melt composition during the erosion. Thus, the resolution rate (increase rate of Fe content in the melt) was calculated using eq. (2). Here, the surface areas of testing portions after testing (total of three specimens) calculated from the erosion ratio of pure molten aluminum under the same assumption as the calculation of the erosion thickness, the diffusion coefficient of Fe in the melt under the testing temperature,³⁵⁾ the melt volume under approximate testing temperature (1000 K),³⁶⁾ the saturation solubility of Fe in the melt based on the calculated Al–Fe binary phase diagram, and the assumptions C = 0 and δ = 0.1 mm³⁷⁾ were used. As a result, the order of the calculated resolution rate was the same as that of the average erosion rate of the total of the three specimens calculated from the erosion ratio, irrespective of the testing time.

Figure 9 shows the relationship between the average erosion rate (total rate of three specimens) calculated from the erosion ratio under the testing time of 2 h and the apparent saturation solubility of Fe in the melts based on the pseudo-binary phase diagrams under the testing temperature. A clear positive correlation that can be approximated by a straight line through the origin was observed, except for the molten Al–17% Si alloy, as shown in Fig. 9. For the effects of Si content in the melt on the erosion behavior, the decrease in the viscosity of the melt and the corresponding increase in the diffusion coefficient have been reported.¹⁶⁾ However, the effects on the diffusion coefficient and thickness of the diffusion layer under the flow condition should be small, indicating that the situation under dC/dt ∝ C₀ in eq. (2) does not depend on the melt composition. These results indicate that the dissolution of the intermetallic layers was caused by the same phenomena described by the Noyes–Whitney–Nernst equation and that the diffusion of Fe in the melt is the rate-controlling step of the dissolution. These also suggest that the difference between the saturation solubility of Fe in the melt and the Fe content reflects the driving force of the dissolution.³⁸⁾ Meanwhile, this dissolution behavior suggests the precondition that both intermetallic and diffusion layers have been already formed. Although the formation processes of the layers should be dominant in the early step of the erosion, the approximate line of the relationship shown in Fig. 9 does not have an obvious positive intercept. This is presumably because those formation processes in the early step of the erosion are also caused under the diffusion-controlled condition of Fe in the melt. In addition, in case the whole of the right side of eq. (2) is constant and does not depend on time, the dissolution rate should be constant irrespective of the testing time. However, the parabolic time variation shown in Fig. 3 was observed. This implies the effects of not only the above-mentioned process in the early stage of the erosion, except for the dissolution of the intermetallic compounds, but also the decrease in the surface area of the testing portions during the testing and increase in Fe content in the melt.

Fig. 9

Relationship between average erosion rate of specimens under testing time of 2 h and apparent saturation solubility of Fe to melts based on pseudo binary phase diagrams under testing temperature.

Figure 10 shows the isothermal section of the Al–Si–Fe ternary phase diagram under the testing temperature. Since the pseudo-binary phase diagrams correspond to the vertical cross sections of the ternary phase diagram, the positions are shown in Fig. 10. The gradients of the straight lines that correspond to the pseudo-binary phase diagram of (Al–3–10.8% Si)–Fe systems were close to those of the tie lines in the intermetallic compound-liquid coexisting areas, as shown in Fig. 10. On the other hand, the gradient of the straight line that corresponds to the pseudo-binary phase diagram of the (Al–17% Si)–Fe system was significantly different from that of the tie lines. Here, the solid–liquid interface during the erosion presumably maintains the state of local equilibrium (the state where the chemical potentials of the solid and liquid are locally equal for each component). For molten Al–17% Si alloy, the distribution of the contents in the pseudo-binary phase diagram may be significantly different from the actual distribution. The plausible example of the actual distribution of the contents is shown in Fig. 10. These insights suggest that the actual saturation solubility of Fe is smaller than its apparent saturation solubility and that the difference between them leaded to the deviation from the approximate line of the relationship shown in Fig. 9.³⁹⁾ For the other melts except for the molten Al–17% Si alloy, the assumptions of the distribution of the contents and the state of local equilibrium based on the pseudo-binary phase diagrams may differ from the actual behaviors. However, the above-mentioned results suggest that the magnitude relationship of the apparent saturation solubility of Fe is almost the same as that of the actual saturation solubility. These findings indicate that the erosion behavior in low-alloy melts can be estimated and compared based on the pseudo-binary phase diagrams, which can be used for applications such as the development of erosion-resistant aluminum alloys when contacted with ferrous materials.

Fig. 10

Isothermal section of Al–Si–Fe ternary phase diagram under testing temperature (gray solid line: tie line).

4. Conclusions

In this study, erosion testing was conducted on iron specimens in various Al-based molten binary alloys, and the correspondences between the constituent phases of the intermetallic layers formed at the contact interface and the thermodynamically stable phases or the pseudo-binary phase diagrams of the Al alloy–Fe systems were examined. Based on the obtained results, the dissolution mechanism of the intermetallic layers and the factors affecting the dissolution behavior were investigated. The following conclusions were drawn from the results:

1. (1)    The erosion ratio of the iron specimens in the molten Al–3% Mn alloy was significantly smaller than that in pure molten aluminum or other Al-based molten binary alloys. Meanwhile, the erosion ratios in the molten Al–Si alloys were greater than that in pure molten aluminum or other Al-based molten binary alloys.
2. (2)    The constituent phases of the intermetallic layers identified by EBSD corresponded to the stable phases based on the equilibrium calculation of the compositions analyzed by EDS and/or the stable phases in the Al alloy-rich portions of the pseudo-binary phase diagrams.
3. (3)    A positive correlation was obtained between the average erosion rate of the specimens calculated from the erosion ratio and apparent saturation solubility of Fe based on the pseudo-binary phase diagrams.
4. (4)    The solid–liquid interface maintained local equilibrium, and the dissolution of the intermetallic layers was caused by the same mechanism described by the Noyes–Whitney–Nernst equation. These results suggest that the diffusion process of Fe in the melt is the rate-controlling step of the dissolution and that the saturation solubility of Fe in the melt is a dominant factor affecting the dissolution.
5. (5)    The dissolution behavior of iron specimens in low-alloy melts can be estimated based on pseudo-binary phase diagrams.

Acknowledgments

This study was supported by the Japan Foundry Engineering Society. This study was also supported in part by the Iron and Steel Institute of Japan (ISIJ Research Promotion Grant).

REFERENCES

• 1)   Y.  Ueda,  M.  Niinomi and  N.  Tohyama: IMONO (J. Jpn. Foundrymen’s Soc.) 48 (1976) 435–440. doi:10.11279/imono.48.7_435
• 2)   K.  Yamamoto,  F.  Yuse,  T.  Nakayama and  F.  Kamikubo: J. Surface Finish. Soc. Jpn. 48 (1997) 449–453. doi:10.4139/sfj.48.446
• 3)   V.  Joshi,  A.  Srivastava,  R.  Shivpuri and  E.  Rolinski: Surf. Coat. Technol. 163–164 (2003) 668–673. doi:10.1016/S0257-8972(02)00693-X
• 4)   A.  Molinari,  M.  Pellizzari,  G.  Straffelini and  M.  Pirovano: Surf. Coat. Technol. 126 (2000) 31–38. doi:10.1016/S0257-8972(00)00530-2
• 5)   A.J.  López and  J.  Rams: Surf. Coat. Technol. 262 (2015) 123–133. doi:10.1016/j.surfcoat.2014.12.023
• 6)  K. Asano, H. Yamada and S. Sugimura: The Proceedings of Mechanical Engineering Congress, 2017 Japan (2017) G0400204. doi:10.1299/jsmemecj.2017.G0400204.
• 7)   M.  Nishi,  Y.  Yamasaki and  K.  Takagi: J. Japan Inst. Metals 66 (2002) 877–880. doi:10.2320/jinstmet1952.66.9_877
• 8)  H. Ike, T. Takagawa, K. Iwashimizu, S. Aso and N. Konishi: Report of the 151st JFS Meeting (2007) p. 37. doi:10.11279/jfeskouen.151_37.
• 9)  H. Ike, T. Takagawa, K. Iwashimizu, S. Aso, N. Konishi and Y. Shobuzawa: Report of the 153rd JFS Meeting (2008) p. 37. doi:10.11279/jfeskouen.153_37.
• 10)   K.  Kusaka and  K.  Murai: Tetsu-to-Hagané 52 (1966) 675–686. doi:10.2355/tetsutohagane1955.52.4_675
• 11)   J.  Ichikawa,  K.  Sudoh,  T.  Ohki and  K.  Kawabe: Denki Seiko (Electr. Furn. Steel) 60 (1989) 343–353. doi:10.4262/denkiseiko.60.343
• 12)   S.-H.  Hwang,  J.-H.  Song and  Y.-S.  Kim: Mater. Sci. Eng. A 390 (2005) 437–443. doi:10.1016/j.msea.2004.08.062
• 13)   U.B.  Kuruveri,  P.  Huilgol and  J.  Joseph: ISRN Metall. 2013 (2013) 191723. doi:10.1155/2013/191723
• 14)  Y. Watanabe, S. Aso, K. Ohguchi, I. Goto, Y. Komatsu, H. Ike and N. Konishi: Report of the 163rd JFS Meeting (2013) p. 12. doi:10.11279/jfeskouen.163_12.
• 15)   Y.  Ueda,  M.  Niinomi and  M.  Sano: J. Japan Inst. Metals 42 (1978) 549–555. doi:10.2320/jinstmet1952.42.6_549
• 16)   Y.  Ueda and  M.  Niinomi: IMONO (J. Jpn. Foundrymen’s Soc.) 50 (1978) 506–511. doi:10.11279/imono.50.8_506
• 17)   M.  Niinomi,  Y.  Suzuki and  Y.  Ueda: J. Japan Inst. Metals 45 (1981) 416–424. doi:10.2320/jinstmet1952.45.4_416
• 18)   Y.  Ueda and  M.  Niinomi: J. Japan Inst. Metals 42 (1978) 543–549. doi:10.2320/jinstmet1952.42.6_543
• 19)   S.  Shankar and  D.  Apelian: Metall. Mater. Trans. B 33 (2002) 465–476. doi:10.1007/s11663-002-0057-7
• 20)   N.  Tang,  Y.-P.  Li,  S.  Kurosu,  Y.  Koizumi,  H.  Matsumoto and  A.  Chiba: Corros. Sci. 60 (2012) 32–37. doi:10.1016/j.corsci.2012.04.015
• 21)   N.  Takata,  M.  Nishimoto,  S.  Kobayashi and  M.  Takeyama: ISIJ Int. 55 (2015) 1454–1459. doi:10.2355/isijinternational.55.1454
• 22)  Jpn. Inst. Met.: Kinzoku Binran, 6th ed., (Maruzen, Tokyo, 2000) p. 571.
• 23)  S. Suzuki: EBSD Dokuhon, 3rd ed., (TSL solutions, Sagamihara, 2009) pp. 52–65.
• 24)   B.  Sundman,  B.  Jansson and  J.-O.  Andersson: Calphad 9 (1985) 153–190. doi:10.1016/0364-5916(85)90021-5
• 25)   M.  Sasabe: Tetsu-to-Hagané 81 (1995) N277–N282. doi:10.2355/tetsutohagane1955.81.6_N277
• 26)   N.  Krendelsberger,  F.  Weitzer and  J.C.  Schuster: Metall. Mater. Trans. A 38 (2007) 1681–1691. doi:10.1007/s11661-007-9182-x
• 27)   J.  Lacaze,  L.  Eleno and  B.  Sundman: Metall. Mater. Trans. A 41 (2010) 2208–2215. doi:10.1007/s11661-010-0263-x
• 28)   M.C.J.  Marker,  B.  Skolyszewska-Kühberger,  H.S.  Effenberger,  C.  Schmetterer and  K.W.  Richter: Intermetallics 19 (2011) 1919–1929. doi:10.1016/j.intermet.2011.05.003
• 29)   K.  Han,  I.  Ohnuma and  R.  Kainuma: J. Alloy. Compd. 668 (2016) 97–106. doi:10.1016/j.jallcom.2016.01.215
• 30)   S.  Lee,  B.  Kim and  S.  Lee: Mater. Trans. 52 (2011) 1053–1062. doi:10.2320/matertrans.M2010422
• 31)   S.  Kores,  M.  Vončina and  B.  Kosec: Metalurgija 51 (2012) 216–220.
• 32)   Y.Q.  Liu,  W.T.  Xu,  D.  Xie and  Z.M.  Li: Mater. Res. Innov. 18 (2014) S2-573–S2-578. doi:10.1179/1432891714Z.000000000489
• 33)   T.  Ohta: Aichi Gakusen University Review 1(2) (2019) 23–33.
• 34)   J.  Song,  X.  Wang,  T.  DenOuden and  Q.  Han: Metall. Mater. Trans. A 47 (2016) 2609–2615. doi:10.1007/s11661-016-3454-2
• 35)   T.  Ejima,  T.  Yamamura,  N.  Uchida,  Y.  Matsuzaki and  M.  Nikaido: J. Japan Inst. Metals 44 (1980) 316–323. doi:10.2320/jinstmet1952.44.3_316
• 36)  Jpn. Soc. Mech. Eng.: JSME Data Book, Heat Transfer, 5th ed., (Maruzen, Tokyo, 2009) p. 323.
• 37)   N.  Shimizu and  N.  Ogata: Bull. Soc. Sea Water Sci. Jpn. 42 (1988) 3–6. doi:10.11457/swsj1965.42.3
• 38)   N.  Nishi and  S.  Kami: J. JFS 70 (1998) 648–654. doi:10.11279/jfes.70.648
• 39)   G.  Miyamoto: Bull. Iron Steel Inst. Jpn. 24 (2019) 10–16.