Regular Article
Growth of Fe2Al5 Phase on Pure Iron Hot-Dipped in Al–Mg–Si Alloy Melt with Fe in Solution
2015 Volume 55 Issue 7 Pages 1454-1459
Details
2015 Volume 55 Issue 7 Pages 1454-1459
We have examined the growth and morphology of the Fe–Al alloy layer (Fe2Al5-η and FeAl3-θ phases) on pure Fe sheets hot-dipped at 750°C in an Al-8.2Mg-4.8Si–Fe (wt.%) and Al–Fe alloy melts which are saturated with Fe where the effect of Fe dissolution from the solid Fe can be eliminated. In both the Al–Mg–Si–Fe and Al–Fe melts, the total thickness of Fe sheets increased during the dipping, indicating that the diffusion of Al from the melt side is dominant in the growth of the Fe–Al alloy layer. The growth rate constant of the η phase layer in the both melts was close to 0.5 while the growth rate in the Al–Mg–Si–Fe alloy melt was approximately one order of magnitude slower than that in the Al-Fe melt. The θ phase layer appeared continuous in the Al–Mg–Si–Fe alloy melt, whereas the layer was discontinuous in the Al–Fe melt. The sluggish growth of the η phase layer in the Al–Mg–Si–Fe alloy melt is discussed.
The hot-dipped Al–Mg–Si alloy coated steel is expected as a candidate material to replace the Zn galvanized steels due to the findings by Tsuru et al.1,2,3) They found that a sufficient sacrificial anodic effect is available in an Al-8.2Mg-4.8Si (wt.%) alloy coating with a fine eutectic microstructure of α-Al and Mg2Si. One of the important issues in the development of the Al alloy coated steels is to obtain sufficient adhesiveness of the coating layer on the steel substrate. Since the brittle Fe–Al intermetallic phase layer is formed in the Al alloy melt as well as in the conventional Al coated steels,4,5,6) it is important to understand and thereby control the formation and growth of the Fe–Al alloy layer.
We have examined the formation and growth of the Fe–Al alloy layer (η-Fe2Al5 and θ-FeAl3 phase layers) on pure Fe sheets hot-dipped in the Al–Mg–Si alloy melt.6) The result demonstrated that the growth kinetics of η phase layer in the case of dipping in the Al–Mg–Si alloy melt is one order of magnitude slower than that in the case of pure Al melt. Besides, the dissolution of Fe in the Al alloy melt occurs, resulting in the thickness loss of the Fe sheets during the dipping. Thus, two competitive reactions, the growth of η phase layer and the dissolution of Fe, occur at the interface between the solid Fe and the liquid Al alloy. The occurrence of Fe dissolution makes it difficult to understand the growth kinetics of the Fe–Al alloy layer. It is generally understood that the solubility of Fe in liquid Al (e.g. about 3 at.% Fe in pure Al liquid at 750°C)5) can provide the driving force for the dissolution of Fe. The effect of the Fe dissolution on the formation of the Fe–Al alloy layer can therefore be eliminated in a saturated solution of Fe in the Al melts, as previously reported.7,8,9)
We have examined the growth kinetics and the morphologies of η and θ phase layers on pure Fe hot-dipped in the Al-8.2Mg-4.8Si (wt.%) alloy melt with a saturated solution of Fe in which the effect of the dissolution of Fe on the growth kinetics are eliminated. The change in the total thickness of Fe sheets with dipping time has also been examined. Based on these results, we discussed the formation and growth process of the η-Fe2Al5 phase layer formed at the interface between solid Fe and liquid Al–Mg–Si alloy.
Fe of 4 N purity was arc melted in argon atmosphere as 40 g button ingot. The ingot was cold-rolled to a sheet with a thickness of about 1 mm, and then annealed at 750°C for 7.2 ks. The pure Fe sheets whose surface was mechanically polished on a #800 emery paper were hot-dipped in an Al-8.2Mg-4.8Si (wt.%) alloy melt with a saturated solution of Fe at 750°C for 2–600 s, followed by water quench. Hereafter the Al alloy melt with Fe saturated is designated as Al–Mg–Si–Fe alloy melt. For preparing the Al–Mg–Si–Fe alloy melt, 400 g of the Al alloy ingots (prepared by casting under argon atmosphere) and 30 g of Fe ingots of 4 N purity were heated to 750°C in an Al2O3 pot and then held for more than 7.2 ks in order to saturate Fe in the melt. Note that the amount of the Fe ingots is much larger than the solubility limit of Fe in the Al alloy melt which was experimentally determined.10,11) The same experimental procedure was carried out to prepare a Fe saturated Al melt (Here after it is designated as Al–Fe alloy melt) by using commercial pure Al and the high purity Fe ingots. The pure Fe sheets were hot-dipped in the Al–Fe alloy melt at 750°C for 30 s, 600 s and 1800 s.
Microstructures were observed using a scanning electron microscope (SEM) at 15 kV and a transmission electron microscope (TEM) at 200 kV. For SEM observation, the sample surface was ion-polished by Cross Section Polisher at 5 kV for over 36 ks. The chemical compositions were analyzed with energy dispersive X-ray spectroscopy (EDS) at 10 kV. A cross-sectional TEM sample was prepared from the interface between the Al alloy coating and the Fe sheet by using the focused ion beam (FIB) pick-up technique12) operated at 30 kV. The thickness of the η-Fe2Al5 and θ-FeAl3 phase layers and the total thickness of the Fe sheets were measured using the SEM images. Since the serrated interface between the η phase layer and the Fe substrate was observed, the thicknesses of the intermetallic phase layers were measured in the following way; (1) the distance between the growing tips of the alloy layer (Fe/η interface) and the bottom (θ/Al interface) was measured, (2) the thickness of the θ phase layer was then measured, (3) the thickness of the η layer was calculated by subtracting the thickness of the θ phase layer from that of the alloy layer. The details on the method used in this study for determining the thickness values are described elsewhere.6)
Figure 1 shows back scattered electron images (BEIs) of the cross sections of Fe sheets dipped in (a) the Al–Mg–Si–Fe alloy melt and (b) in the Al–Fe melt. In these images, broken lines indicate the positions of the initial Fe/Al interfaces determined by a measurement of the Fe sheet thickness before dipping. The broken lines indicate that an increase in the total thickness of Fe sheets by dipping and the formation of Fe–Al alloy layer (gray colored region) are much more pronounced in the Al–Fe alloy melt than in the Al–Mg–Si–Fe alloy melt.
BEIs showing the cross section of Fe sheets hot-dipped in (a) Al–Mg–Si–Fe alloy melt and (b) Al–Fe alloy melt for 600 s.
Figure 2 shows the change in the total thickness of Fe sheets dipped in the Al alloy melts with dipping time. The differences from the initial thickness (before dipping) are plotted on the vertical axis. In the Al–Mg–Si–Fe alloy melt, the total thickness slightly increases with increasing dipping time. The thickness gain is about 30 μm after 600 s. In the Al–Fe alloy melt, the total thickness remains unchanged till 30 s, but then substantially increases with increasing dipping time. The total thickness reaches to a value more than 1300 μm after 1800 s.
Change in the total thickness of the Fe sheets dipped in Al alloy melts with dipping time.
Figure 3 shows a TEM bright field image and selected area electron diffraction (SAED) patterns of the Fe–Al alloy layer in the Al–Mg–Si–Fe alloy melt for 10 s. The SAED patterns were obtained from the corresponding regions of (b) and (c) in the TEM image (Fig. 3(a)). The SAED patterns represent the incident beams parallel to [110]θ (Fig. 3(b)) and [110]η (Fig. 3(c)), which confirms the constituent phases of FeAl3-θ (mC102) and Fe2Al5-η (oC24) phases in the Fe–Al alloy layer in the Al alloy melt. The structures identified are the same as reported in the Al–Mg–Si alloy melt.6) A small amount of Si-rich particles are formed in the η phase layer. No other intermediate phases were observed at the η/α-Fe interface.
(a) TEM bright-field image and selected electron diffraction patterns for (b) FeAl3-θ and (c) Fe2Al5-η phases in the Fe–Al alloy layer in pure Fe sheet hot-dipped in the Al–Mg–Si–Fe alloy melt for 10 s.
Figure 4 shows a change in the morphology of the η phase layers formed on pure Fe sheets hot-dipped in the Al–Mg–Si–Fe alloy melt and in the Al–Fe alloy melt. In both the Al–Mg–Si–Fe alloy and Al–Fe alloy melts, the η phase layer exhibits saw-tooth morphology on the Fe side, as also reported in the literatures.4,7) The morphology is due to the directional growth of η phase grains perpendicular to the solid/liquid interface. It is obviously seen that the layer is much thinner in the Al–Mg–Si–Fe melt than in the Al–Fe alloy melt.
BEIs showing Fe2Al5-η phase layer formed on pure Fe sheets hot-dipped in (a, b) Al–Mg–Si–Fe alloy melt and (c, d) Al–Fe alloy melt for (a, c) 30 s and (b, d) 600 s.
Figure 5 compares the morphology of θ phase layers formed in the both alloy melts. In both the Al alloy melts, thin θ phase layers are present at the interfaces between the η phase layer and the Al alloy melts. In the Al–Mg–Si–Fe alloy melt (Figs. 5(a) and 5(b)), the θ phase layer appears continuous with a layer thickness of about 3–5 μm. On the other hand, in the Al–Fe alloy melt, the θ phase layer shows a discontinuous layer morphology and its thickness is somewhat inhomogeneous (Fig. 5(c)), in comparison with that in the Al–Mg–Si–Fe alloy melt. The discontinuity and the inhomogeneous thickness of the layer appear more significant after 600 s (Fig. 5(d)). The discontinuity of the θ phase layer indicates that the local interface between the η phase and the Al–Fe liquid existed during dipping.
BEIs showing FeAl3-θ phase layer formed on pure Fe sheets hot-dipped in (a, b) Al–Mg–Si–Fe alloy melt and (c, d) Al–Fe alloy melt for (a, c) 30 s and (b, d) 600 s.
Figure 6 shows the change in the thickness of η and θ phase layers in the both Al alloy melts with dipping time. The thickness of the η phase layer continuously increases with increasing dipping time (Fig. 6(a)). Its growth kinetics in the Al–Mg–Si–Fe alloy melt is one order of magnitude slower than that in the Al–Fe melt. The growth rate constants in both melts are close to 0.5, which indicates the growth kinetics on the η phase layer follow a parabolic law. The growth rate constant was also reported in the previous report.7) The thicknesses of the θ phase layers in both the Al alloy melts are approximately 3 μm after 2 s and then slightly increase with increasing dipping time (Fig. 6(b)). The mean thicknesses of the θ phase layers are similar, about 5 mm after 600 s, in both the Al alloy melts, but the scattering in the thickness is obviously larger in the Al–Fe alloy melt than in the Al–Mg–Si–Fe alloy melt. The different scattering in the thickness of the θ phase layers is associated with the different morphologies of the layers, as described before. The growth rate constants of the θ phase layers are less than 0.2.
Changes in the thicknesses of (a) η phase and (b) θ phase layers in the Fe sheets hot-dipped in Al alloy melts with dipping time at 750°C.
Figure 7 shows the composition profile across the Fe–Al alloy layer formed on the Fe sheet dipped in the Al–Mg–Si–Fe alloy melt. The concentration of Fe in the Al alloy melt is approximately 2.0 at.% with a deviation of 0.6 at.% and keeps almost constant around the interface between the liquid and the solid phase (θ phase layer). In the thin θ phase layer, the Fe concentration of 24 at.% and the Si content of 2–3 at.% were detected. In the thick η phase layer, the Fe content is almost constant (~28 at.%), while the Al content is fluctuated probably due to the existence of Si-rich particles in the η phase. The concentration of Fe is ~99% in the solid Fe, even in the vicinity of the η/Fe interface.
Concentration line profiles across the Fe–Al alloy layer formed on pure Fe sheet dipped in the Al–Mg–Si–Fe alloy melt at 750°C for 600 s.
The present results demonstrate that the growth of the η phase layer can be substantially suppressed by addition of Si and Mg to the Al–Fe alloy melt, even under the condition of suppressed Fe dissolution in the Al melts, as well as under the dissolution of Fe into Al melts (Fe-free Al–Mg–Si alloy and pure Al melts).6) The sluggish growth of the η phase layer has been observed in the Al–Si binary alloy4,13) and Al–Si–Cu ternary alloy melts14) as well, which indicates that the same mechanism would work for the sluggish growth, regardless of added third elements (Fe, Mg and Cu) in the Al–Si alloy melt. Based on the present results, the mechanism of the sluggish growth of η phase in the Al–Mg–Si alloy melt will be discussed below.
One of its possible mechanisms is that the added Si would change the free energies of η phase and/or Al liquid phase, resulting in the reduced driving force for the formation of η phase in liquid Al. In the η phase, several atomic percent Si can be solved at elevated temperatures, as shown in the Fe–Al–Si ternary phase diagram.15,16) Our composition analysis revealed that the η phases in the Al–Mg–Si–Fe alloy melt contains approximately 2–3 at.% Si in solution but no detectable Mg (Fig. 7). The Si solubility suggests the solute Si would reduce the free energy of the η phase. On the other hand, the additions of Si and Mg reduce the melting point of liquid Al,17) which indicates that the free energy of the liquid phase would be reduced by dissolved Si. The consideration cannot determine the change in the driving force for the formation of η phase in liquid Al by Si addition. Thus, the suppressed growth of the η phase might not be explained in terms of free energy of formation.
Another possible mechanism is the controlled Al diffusion from liquid Al alloys into the η phase. In previous studies,13,18,19) it was supposed that the retarded Al diffusion in the η phase might be associated with the occupation of vacancy sites in the oC24 structure by solute Si atoms. This assumption would be a possible reason for the sluggish growth of η phase layer, because of 2–3 at.% solute Si in the η phase formed in the Al–Mg–Si–Fe alloy melt (Fig. 7).
However, the microstructural observation can suggest that another considerable factor contributing to the growth of the η phase layer is the continuity of the θ phase layer formed in the Al–Si–Mg–Fe alloy melt. This mechanism is closely associated with the interfacial reaction between solid Fe and liquid Al. The formation process of the θ and η phase layers at the interface between pure Fe and Al alloy melts (the Al–Mg–Si–Fe alloy or the Al–Fe alloy) is schematically shown in Fig. 8. Since the dissolution of Fe is suppressed by saturated Fe in the Al melts, the reaction is considered to start with the formation of θ phase layer at the interface between solid Fe and liquid Al/Al alloy with a saturated solution of Fe (Figs. 8(a) and 8(d)). The η phase can be formed either through the liquid-solid reaction between the liquid Al and solid Fe (Fig. 8(b)), or through a solid-solid reaction between the θ phase and solid Fe (Fig. 8(e)). In the Al–Fe alloy melt, the θ phase layer was found to be discontinuous, which leads to the direct diffusion of Al from liquid Al into the η phase, resulting in the fast growth of the layer towards both sides of the liquid Al and the solid Fe (Fig. 8(c)). On the other hand, in the Al–Mg–Si–Fe alloy melt, the θ phase layer was found to be continuous, which causes the inbound Al diffusion through a solid-solid reaction between the θ phase and Fe substrate (Fig. 8(f)), resulting in the slow growth of the η phase layer. One can therefore assume that the continuous θ phase layer may play an important role of the diffusion barrier to avoid the inbound Al diffusion into the η phase through the solid-liquid interface.
Schematic illustrations showing the interfacial reaction process between solid Fe and liquid Al with a saturated solution of Fe: (a–c) Al–Fe alloy melt, (d–f) Al–Mg–Si–Fe alloy melt.
The above-mentioned reaction process represents that the η phase grows towards both liquid Al and solid Fe sides, which follows a parabolic law (Fig. 6). This result suggests that the growth of the η phase would be controlled by inbound Al diffusion from liquid Al. The front of η phase growing toward Fe side (around the interface between η phase and Fe) would not be in local thermodynamic equilibrium, which is represented by the discontinuous concentration profile across the η/Fe interface (Fig. 7) and its irregular interface morphology (Fig. 4). The nearly constant concentration of Fe within the η phase (Fig. 7) might be related to a limited composition range of the η single phase.5)
The comparison with our previous result of the Fe-free Al–Mg–Si alloy melt6) demonstrates that the θ phase with approximately 2–3 at.% Si in solution (no detectable Mg) exhibits a continuous layer morphology, regardless of dissolved Fe in the Al–Mg–Si alloy melt. This suggests that the solute Si might contribute the change in the morphology of the θ phase layer. In the θ phase in equilibrium with liquid phase in an Al–Si–Fe ternary system, several atomic percent Si can be solved at elevated temperatures.15,16) Thus, the Si in solution would stabilize the θ phase, which may enhance the formation of θ phase in the liquid Al–Mg–Si alloy with a supersaturated Fe, resulting in the continuous layer morphology of the θ phase on pure Fe (Fig. 8(d)). However, whether the added Si (or Mg) can increase a driving force for the formation of θ phase in the liquid Al–Fe alloy is still unclear, since a change in the free energy of the liquid phase by addition of Si and Mg has not been determined. In order to fully understand the formation of the continuous θ phase layer in the Al–Mg–Si alloy melt, further studies for free energy of formation in an Al–Mg–Si–Fe quaternary system are required.
We have examined the growth and morphology of Fe–Al intermetallics layers (Fe2Al5-η and FeAl3-θ phases) on pure Fe sheets hot-dipped in an Al-8.2Mg-4.8Si (wt.%) alloy and pure Al melts with a saturated solution of Fe in which the effect of the dissolution of Fe on the growth kinetics are eliminated. Based on these results, the growth of η phase layer was discussed in terms of the interfacial reaction between solid Fe and liquid Al alloys. The conclusions are as follows:
(1) The Fe–Al alloy layer formed in the Fe saturated Al–Mg–Si alloy (Al–Mg–Si–Fe alloy) melt consists of a continuous θ-FeAl3 phase layer on liquid Al alloy and a large η-Fe2Al5 phase layer on solid Fe. The growth kinetics of the η phase layer in the Al–Mg–Si–Fe alloy melt is approximately one order of magnitude slower than that in the Fe saturated Al melt (Al–Fe alloy melt). In both Al alloy melts, the growth of η phase towards liquid Al and solid Fe sides follows a parabolic law. The θ phase layer exhibits much slower growth kinetics in both Al alloy melts. These results demonstrate that the addition of Si and Mg into the Al melt significantly prevents the growth of η phase under the condition of suppressed Fe dissolution.
(2) The θ phase layer appears the continuous layer morphology in the Al–Mg–Si–Fe alloy melt, which is quite different of the discontinuous θ phase layer in the Al–Fe melt. The continuous layer morphology of the θ phase was observed in the Fe-free Al–Mg–Si alloy melt as well.6) The continuous θ phase layer would act as the diffusion barrier against Al to solid Fe, which contributes to the sluggish growth of η phase in the Al–Mg–Si–Fe alloy melt.
The authors would like to thank the Ministry of Education, Sports, Culture, Science and Technology of Japan for financial support through Element Science Technology Project “Development of Hot-Dipped Aluminum Alloy Coated Steels”.
