Aluminum Components for Automobiles:
Casting Low-Cost, High-Quality Parts

o meet the needs of today's auto industry, manufacturers have begun producing aluminum components by pouring the metal into molds when it is already partially solidified. The lightweight precision parts produced cost less than those made by forging and machining solids and are stronger and more reliable than those made by conventional casting of liquids. "Semisolid processing" became possible 25 years ago, when an MIT graduate student found that stirring molten metal while it solidifies produces a semisolid that can be cast even when it is solid enough to be handled by robotic equipment. Now the MIT researchers are using industrially important aluminum alloys and new experimental techniques to clarify exactly why the process works and how to make it work better. As expected, the more solid the material is, the shorter the distance it flows before stopping--not good behavior for filling a mold. However, partially solidified samples flow farther if they have been formed by cooling the molten metal slowly and stirring it vigorously. Microstructural analysis shows why. Slow cooling and long stirring produce particles that are large and spherical, while fast cooling and short stirring produce small, jagged particles that tend to stick together. As a result, the larger particles slide by each other more easily than the smaller ones do, producing a semisolid with less resistance to flow. High-speed videos show that the semisolid advances smoothly and evenly, producing parts with a consistent microstructure, smooth finish, high reliability, and constant properties.

Despite those and other advantages, semisolid processing did not take off until the mid-1990s, when the automotive industry began calling for parts that were lightweight, low-cost, reliable, and long-lived. Cast or forged components made of iron or steel began being replaced by components of aluminum alloys manufactured by semisolid processing. The advantages of semisolid processing over traditional methods have proved remarkable. Castings of semisolid rather than molten metal form quickly and have lower cooling requirements. Semisolids can be handled by robotic arms and yet still flow well enough to fill intricate molds--in fact, better than molten metals can. Less gas is entrapped, fewer voids form, the surface finish is smoother, and less shrinkage occurs. The parts produced have more consistent and better mechanical properties and closer tolerances; and they need little if any additional shaping after they are formed, saving manufacturing time and expense.

Each year, companies in the United States, Japan, and Europe now produce or use millions of parts made by semisolid processing, and the number is growing quickly. However, as often happens, the commercial adoption of the new process has far outpaced the development of a detailed understanding of how it works. For example, how do the length of stirring, the rate of cooling, and other processing conditions affect the microstructure of the semisolid material? How does that microstructure affect how the material flows? And how does that flow behavior affect the products formed? An Energy Laboratory team including Professor Flemings, Anacleto de Figueredo, and Yusuf Sumartha has been performing rigorous experiments to answer such questions. While the focus is on expanding fundamental understanding, the results will have practical implications, providing manufacturers better control over the process and the products it yields.

To understand why stirring changes the behavior of semisolids, one must consider the microstructure of solidifying metals. Nearly all metals and alloys of commercial importance solidify "dendritically." As the liquid begins to solidify, tiny pine-tree-shaped particles called dendrites form (see the figure below). These jagged particles do not slide by one another easily, so the semisolid material does not flow well. Stirring the liquid as it solidifies breaks up the dendrites as they form, tearing their branches off. The particles are smoother and can slide by each other. As a result, the semisolid flows more easily, even when its solid content is relatively high. Professor Flemings likens the process to making ice cream. If you combine all the ingredients for ice cream and place the mixture in a freezer, the material formed will contain big ice crystals--not a very palatable product. But if you stir the mixture while it is freezing, you break up the ice crystals as they form. The ice cream remains smooth and creamy, even when as much as 70% of it is solid.

Previous studies have looked at the flow behavior of semisolids, but their approach typically has not accurately reflected the industrial situation. For example, the studies tended to involve small quantities of semisolid material. The bulk quantities used in a factory might behave quite differently. And they generally involved measuring the viscosity of the semisolids--a property that does not translate directly into how well a semisolid will flow into the little spaces inside an industrial mold.

Professor Flemings and his colleagues therefore use a different approach--one that involves both a larger amount of metal and a different method of quantifying flow behavior. They melt an ingot of an aluminum alloy in a crucible. They then cool the molten metal while stirring it at a controlled speed between 0 and 1000 rpm. When the metal is partially solidified, they use a partial vacuum to suck some of it up a silica tube and then measure how far it travels before it solidifies. That technique, specially adapted for use with semisolid materials, is a simple but effective means of establishing "fluidity," a foundry concept that describes the ability of a metal to flow before being stopped by solidification.

In the experiments, thermocouples placed in the metal directly measure its temperature. Videos taken at 40,000 frames per second track how fast the semisolid moves up the tube, and quantitative metallographic analysis shows the microstructure of the specimens collected in the tube. (Although the specimens are solid, their microstructure is similar to the material in semisolid state: solidification occurs so quickly in the tube that the microstructure has no time to change.) Based on all the experimental evidence, the researchers relate fluidity to the microstructure of the semisolid, and the microstructure to processing conditions such as cooling rate, cooling time, and "fraction solid" (the fraction of the semisolid that is in solid rather than liquid form).

With so many interacting factors involved, identifying conditions that increase the fluidity of the semisolid is not straightforward. The figure below summarizes some of the experimental results. The curves show how fluidity changes as the fraction solid changes, with each curve representing a different cooling rate. Not surprisingly, as the semisolid becomes more and more solid, its fluidity decreases, regardless of the cooling rate. For a given fraction solid, a slower cooling rate produces higher fluidity. Thus, if the material is about 14% solid (by volume), cooling it at 1.7^oC per minute produces a semisolid that travels about 73 cm up the tube, while cooling it at 1.0^oC produces a semisolid that travels about 80 cm. The explanation for that trend is clear. If the molten metal is cooled more slowly, it will take longer for it to reach a given fraction solid. Assuming that stirring is continuous, the longer processing time translates into a longer stirring time, thus fewer dendrites and higher fluidity. (Stirring the solidifying ice cream longer makes it more creamy.) Videos taken with the high-speed camera show that even at a high fraction solid, the well-stirred semisolid material moves up the experimental tube smoothly and consistently.

Almost more interesting are the microstructures that occur under the different processing conditions. For example, the figure below shows the microstructures of two specimens collected at the same fraction solid. The bottom one was cooled more slowly, thus stirred longer, than the top one. The lower cooling rate and longer stirring time produce larger particles. Yet the fluidity tests show that longer stirring increases fluidity. Several factors explain that seeming contradiction. For one thing, the larger particles are spherical, while the smaller ones are irregular in shape, thus less able to slide by one another. In addition, the smaller particles have a higher surface area and are more likely to interact and stick together than the larger particles are. The result is irregular agglomerates that move less easily than the large spherical particles do.

In those experiments, the researchers completely melted the aluminum alloy and then worked with the semisolid as it formed during cooling. But that approach replicates only one procedure used by industry. Parts manufacturers may heat solid ingots made from stirred liquids to produce semisolid materials to cast. To explore that situation, the researchers performed experiments following the same procedure. They heated solid ingots specially prepared by the Alumax Company until they were partly solid and partly liquid. They then stirred the mixture for variable periods of time before sucking samples up into the experimental tube. 

The results are quite different from those of the tests starting with molten metals. The researchers were able to make a semisolid shape that was solid enough to be handled without deforming but soft enough to form in a mold. However, for a given fraction solid, the specimens collected in the experimental tube have substantially lower fluidity than do those produced directly from liquids. Stirring the reheated semisolids makes them flow more freely, but even after a full hour of stirring they are not as fluid as the materials produced by stirring molten metal continuously as it cools. Micrographs show that the original ingots contain relatively fine, rounded particles. On remelting, those particles tend to form three-dimensional, entangled strings. The strings are still present after an hour's stirring, and some of them are as long as the experimental tube is wide.

Thus, starting with remelted solid ingots is not as good as starting with fully molten metal. However, the drawbacks of using remelted solids are easily offset by the convenience and cost savings of not having to deal with large quantities of molten metal. While homemade ice cream may have a nicer taste and consistency than store-bought, most people choose to buy their ice cream in the store and let it melt a bit so it becomes smooth and creamy.

Experimental equipment now being designed and built by James Yurko will provide the research team with an even closer look at how semisolids flow. Instead of sucking the stirred semisolid up a tube, the new experimental apparatus will push it through a small hole, much like pushing toothpaste out of a tube. As a result, the semisolid will move about ten times faster than in the previous experiments, achieving velocities close to those that prevail in important industrial processes such as die casting. In addition, the new equipment will be highly instrumented. The direct measurements of velocity, pressure, and other important characteristics will provide the researchers with a much improved fundamental understanding of how semisolids flow during industrial processing.
 

Merton C. Flemings is Toyota Professor of Materials Processing. Anacleto de  Figueredo is a research staff member and James Yurko is a PhD candidate in  the Department of Materials Science and Engineering. Yusuf Sumartha  received his SM from the same department in fall of 1997. This research was  supported by the University Research Consortium of the Idaho National  Engineering and Environmental Laboratory and by the Alumax Corporation.  Further information can be found in references.