Understanding Segregation to Predict Solidification
Solidification paths may be calculated from segregation equations, and may be used to pinpoint optimum alloy compositions for your castings.
Geoffrey Sigworth, GKS Engineering Services, Dunedin, Florida
(Click here to see the story as it appears in June's Modern Casting.)
This article is the last in a series on solidification in aluminum castings. While the series focuses on solidification principles in aluminum alloys, many can be applied to other metals, as well.
When a metal is liquid, it is homogeneous. That is, the metal properties (especially composition) are the same everywhere. However, during freezing there is a redistribution of alloying elements and impurities called segregation. Segregation has a significant influence on casting quality. Understanding the process is useful for the casting engineer or metallurgist.
The two types of segregation are microsegregation and macrosegregation. Microsegregation is the variation in composition on a very small scale: between dendrites and dendrite arms.
The Al-Cu alloy system will be used to illustrate how segregation occurs. Researchers have examined this alloy system in great detail by researchers. It forms the basis for the high strength family of 2xx casting alloys.
Figure 1 shows the complete Al-Cu phase diagram. A number of intermetallic compounds form in this system. Aluminum casting facilities are concerned with the aluminum-rich portion of the figure and the formation of the ϴ (Al2Cu) phase. The relevant portion of the figure is shown in Fig. 2.
If an alloy containing 4.5% copper is poured and held near the eutectic temperature, the casting will be in the single phase field corresponding to solid aluminum. This is indicated by the red box in Fig. 2. According to the phase diagram, this alloy should be a single phase—aluminum with copper in solid solution.
Looking at a sample from a casting of this alloy, it is found to contain a significant amount of eutectic Al2Cu phase. This eutectic should not be present, according to the phase diagram. So, it is called a non-equilibrium eutectic.
Also, if the distribution of copper in a sample of the casting is studied by microprobe analysis in a SEM, one finds that the copper content in the aluminum phase varies. In the center of dendrite arms, which corresponds to the first solid, the copper content is low. Moving toward the outside of the arms, which corresponds to metal freezing later, the copper content increases. An example of this type of measurement is shown in Fig. 3.
Solidification scientists have studied this phenomenon for many years and have offered models to explain microsegregation. One can now calculate the amount of non-equilibrium eutectic according to a model proposed by Brody and Flemings (B-F). The B-F equations may be used to calculate the composition of liquid during solidification, and the fraction of solid at which the eutectic forms. The B-F works well for the elements commonly found in aluminum castings and for all but extremely fast freezing rates. (The model works well in other metal systems for most elements. However, a correction factor is needed for elements that diffuse rapidly—for example, carbon dissolved in iron or steel castings.)
Macrosegregation is the redistribution of solute elements on a larger scale. In net-shaped castings this is caused by the motion of fluid inside the casting and usually is related to feeding. Macrosegregation is most easily seen in copper-containing alloys. An example is shown in Fig. 4.
This X-ray shows a section of a cylinder head cast in 319 alloy. Two risers are visible at the top. Both show shrinkage cavities. Underneath the risers are dark bands in the casting, where copper-rich liquid has been “sucked into” the casting.
Risers are placed in the mold to feed liquid to the solidifying metal, whose volume shrinks by 5-7% during solidification. When the risers are not sufficiently large, they will feed the casting late in the solidification when the remaining liquid is enriched in copper (and other solute elements).
Macrosegregation of this sort also can develop in Al-Si casting alloys, but it is more difficult to see this in casting X-rays. (The density difference between aluminum and silicon is much smaller than between aluminum and copper.)
Calculating Solidification Paths
Equations for microsegregation during solidification can be used to calculate solidification paths for casting alloys.
The easiest way to proceed is to assume the distribution coefficients for the binary alloy systems can be used in ternary alloy systems. The starting point for our calculations is the aluminum rich corner of the Al-Si-Fe ternary. Eleven different intermetallic compounds have beenidentified in this system. Four of them occur in the aluminum-rich portion of the ternary presented in Fig. 5. They are:
- FeAl3, which is found in the Al-Fe binary and in alloys low in silicon.
- α-FeSiAl, which has a composition close to Fe2SiAl8.
- β-FeSiAl, which is usually represented by the composition FeSiAl5.
- δ-FeSiAl, which has the composition FeSi2Al3.
The compound which is of most concern here is the β phase. This is the intermetallic compound normally observed in commercial castings.
The calculated solidification paths for an AA309 alloy having 5% silicon and various iron contents have been calculated. (This alloy also has 1.2% copper and 0.5% magnesium, but this is ignored in the calculation.) To simplify presentation of the results, only the phase boundaries are shown. Also a detailed section from Fig. 5 is taken. The result is shown in Fig. 6 for alloy iron contents of 0.3 and 0.6%. Two segregation curves are given for each case. The lower (red) curve is for a solidification time of 10 seconds. The upper (blue curve) is for a longer freezing time of 1,000 seconds.
From this result, it can be seen that 0.3% iron is a borderline case for this alloy. In rapidly solidified parts of a casting there should be no primary β phase, only a ternary eutectic according to this reaction:
Liquid → Al(solid) + Si(solid) + β
At slower solidification rates, however, primary β should form. At higher iron contents (e.g., 0.6%) primary β forms before any Al-Si eutectic, regardless of the freezing rate.
This alloy was studied by the Center for Advanced Solidification Technologies (CAST) researchers in Australia. They found that casting defects were associated with iron contents that produced primary β. However, when they switched to a higher silicon version of the same alloy, the defects went away. The reason for this behavior may be seen by considering Fig. 7. Similar calculations are made for the same two iron contents. In this alloy the higher iron content (0.6%) becomes the borderline case. Therefore, 9% silicon alloy can tolerate twice the iron content of the 5% silicon alloy.
In exactly this manner, the CAST researchers calculated solidification paths for numerous silicon contents. In this way they derived a map of safe iron contents for their casting (Fig. 8).
The results of related research, conducted by Caceres and co-workers, should also be considered. They produced castings in a number of alloy compositions and measured mechanical properties. Some of their results
are shown in Fig. 9, which illustrates the importance of microsegregation during segregation, and how higher silicon contents may be used to advantage in Al-Si based alloy castings.
The tensile strengths for castings heat treated to the T6 temper are shown on a quality plot (Ultimate Tensile Strength [UTS] versus the log of elongation).
The red lines show constant values of quality index (in MPa). The blue arrows indicate the change in alloy composition. For example, iron was added to alloy 1 to obtain alloy 2. The result was a significant loss in casting quality—about 120 MPa according to the quality index.
Silicon was added to alloy 2 to obtain alloy 3. Nearly all of the lost quality was regained by increasing the silicon content from 4.5 to 9%.
A similar result was found going from alloys 1 → 6 → 7, except in this case copper was added along with the iron.
The loss in quality with the combined addition of iron and copper was larger—about 200 Mpa—but that loss was regained by increasing the silicon content.
By contrast, when copper was added by itself; in the alloys 1 → 4 → 5, only a small loss of quality was found in alloy 4.
This article was based on Paper 13-1224, which was presented at the 117th Metalcasting Congress.