Semi-solid Processing of Alloys

The original semisolid forming process, developed at MIT in 1972, involved stirring an alloy during solidification to produce a slurry of spheroidal primary particles in a liquid matrix, which was then injected directly into a die to produce a solid component. This was termed “rheocasting.” Subsequently, it was found more convenient to solidify the slurry completely during the continuous casting of an electromag netically stirred strand, which was then cut into slugs for partial remelting back into semisolid billets on demand. These could be loaded into a diecasting machine in this state for injection into the die. This alternative process route is called “thixoforming,” and until recently, it was the preferred industrial process. For this reason, the microstructure developed during the reheating and melting for thixo forming will be considered first. Experimentally, it has been found that the most effective fraction solid f _s for thixoforming, lies between 0.5 and 0.6. Below this range, the semisolid slug becomes too soft to support its own weight and sags during remelting; above this range, it is too stiff to flow readily and fill the die. However, the slurry technologies used in rheocasting typically operate at lower fraction solids and rely on the ability to pour the semisolid alloy much like a liquid (see Chap. 4). It has been observed in practice that the fraction solid is in fact a critical factor for effective thixoforming, and therefore, both good temperature control and lack of sensitivity of f _s to small temperature variations of the alloy are essential to efficient manufacturing.

It has been understood for some time that the flow behavior of semisolid metal slurries, and the properties of parts shaped from the slurries, depend both on the fraction solid and on the size, distribution, and morphology of the solid particles within the liquid matrix. The fraction solid f _s is essentially determined by the temperature for a given alloy; the other features, however, are a function of the history of preparation and it is known that slurry flow is enhanced by achieving fine spherical particles. Fraction solid (f _s) can be determined from phase diagrams for simple alloy systems if equilibrium may be assumed, or from direct measurement on rapidly quenched specimens, using metallographic techniques on sectioned surfaces, such as line intercept: f _s = L _a, where L _a is the total intercept length in the a particle phase per unit length of test line. Likewise, the average particle size (assuming spheres of constant diameter d) may be obtained by the measurement: d = L _a/N _a, where N _a is the number of grains per unit area of sectioned surface. It should be noted that the fraction solid obtained from phase diagrams is strictly a weight fraction and only coincides with the volume fraction obtained from metallographic measurements above when the densities of liquid and solid are considered equal and shrinkage is therefore ignored. Other techniques have been used to obtain these quantities, but are indirect, requiring calibration, and are probably less reliable. However, see in situ X-ray microtomography in Sect. 3.4, where the measurement is carried out in the semisolid condition to avoid quenching artifacts.

In a recent review of coarsening, Flemings [19] states that this process refers to the growth “of solid regions of low curvature at the expense of regions of higher curvature,” and this includes “the growth of larger particles or dendrite arms with the simultaneous dissolution of smaller particles or arms (so-called ‘ripening’), the filling of spaces between particles or dendrite arms (‘coalescence’) and the breakup of dendrites (‘dendrite multiplication’).” This is a view adopted in the present chapter. The driving force is of course always the reduction of total solid–liquid interface area and the reduction in the associated interfacial energy, and the general mechanism by which this is achieved is diffusion of solute atoms through the liquid from concentration gradients established between regions of high and low curvature. Owing to the complex geometries that may exist in coarsening semisolid systems with competing fluxes between different particles, which can change with both time and place, simple kinetic equations to describe the overall coarsening process is probably not possible. However, where the geometry of the solid may be clearly described, for instance in terms of dendrite arms formed early in solidification, simple kinetics can be derived. More complex geometries may develop later during coarsening that are more difficult to provide with an adequate geometrical description.

The semisolid state in a metallic alloy can be obtained either during solidification from the liquid state, or during partial remelting from the solid state. Metal forming during solidification (rheocasting) has been studied in the early stages of the development of semisolid processing by Flemings and his coworkers, but it was not the main forming method until very recently with the development of the new rheocasting (NRC) process. However, solidification of an alloy with mechanical, passive, or electromagnetic stirring was extensively used to produce globular microstructures. Therefore, knowledge of the rheological properties of alloys subjected to stirring during solidification is of great importance in comparison with non-stirred alloys on the one hand and with partially remelted alloys on the other hand. Metal forming was thus mainly carried out after partial remelting: thixoforming (thixocasting or thixoforging) is then concerned with solid fractions, which can vary in a quite large range, from 10 to 20% in thixomolding to much larger values in thixoforging. Study of the rheological behavior of alloys during such a treatment is therefore also important for better forming conditions and improved properties of the formed components. The alloys studied from a rheological point of view belong essentially to three categories: the model alloys like Sn–Pb with a low melting temperature for which the experiments are relatively easy, the aluminum and magnesium alloys, which are the main thixoformed alloys, and the alloys with a high melting temperature for which only feasibility tests have been carried out.

This chapter will be divided into two main parts. The behavior of alloys during partial solidification will be considered first but only alloys with a globular morphology will be examined. In this case, the alloy is considered as a homogeneous medium and therefore, the viscosity only (and sometimes the fluidity) has been determined. Alloys during partial remelting will thereafter be considered. In this case, the solid fraction can vary in large proportions so that the two approaches mentioned above will be detailed. In each part, the experimental methods will be examined first and then, the experimental results will be presented. It is necessary to mention here that only the most representative results will be presented and not all of the results in the literature. In addition to these two main parts, a comparison between partially solidified and partially remelted alloys will be performed. This comparison is important in view of the latest industrial developments of the rheocasting processes. Finally, the question of yield stress will be addressed.

Computational fluid dynamics (CFD) can be used to predict die filling. However, input parameters must be obtained from rheological experiments (see Chap. 6. In some cases, the data that has been used is from steady state experiments, where the material has been maintained at a particular shear rate for some time. In cases where the process involves taking material from rest into the die in a fraction of a second, it is difficult to see how this can be appropriate. The material changes viscosity by several orders of magnitude in that fraction of a second [23, 32]. Observations of transient rheological behavior under rapid changes in shear rate are therefore more relevant for input data for modeling. Modeling of semisolid processing is generally continuum modeling, where the macroscopic behavior is predicted with the internal structure represented by a few internal variables. Continuum modeling for semisolid can be categorized into onephase or two-phase and as finite difference or finite element. In one-phase modeling, a single non-Newtonian equation of state for the viscosity takes into account the interrelationship between the liquid and the solid through the way in which they behave when they are in combination. In two-phase modeling, each individual phase (the liquid matrix or the solid particles) is characterized with separate equations. The two-phase approach is more complicated, requires significantly higher computational time and much higher effort on obtaining the experimental parameters for input. It is, however, more physically realistic and does allow for important (and undesirable) phenomena, such as liquid segregation to be modeled. Liquid segregation is where liquid gathers in one part of the component in preference to another during processing. Figure 7.1 shows a classic but extreme example. The material is being forced vertically upwards into a die and has to flow around a corner to reach the end of the die. The sharp corner (the dotted line in the inset shows the die shape) causes the liquid to separate from the solid and to gather in one place. In the micrograph, the quenched liquid is dark gray in comparison with the solid. The die has not filled properly. Liquid segregation is deleterious, leading, for example, to inhomogeneous mechanical properties after solidification.

Although semisolid forming is far from replacing conventional casting and forging processes, it has become interesting for the production of specific parts for various industrial sectors. In order to avoid time-consuming trial-and-error experiments to design the mould or the die and to determine the optimum forming conditions, numerical simulation is now extensively developed. It requires an important research activity both to improve the simulation methods for better accuracy and reduced computer time and to develop models that reproduce at best the behavior of the material during the forming operation. In Part II, the experiments that have been developed to determine the rheological behavior of alloys in the semisolid state have been described together with the representative results of these experiments. The models which have been proposed to account for this behavior have also been detailed. The behavior of this type of material is now much better understood but there are still debates concerning particular aspects, such as the existence of a yield stress. In addition, many researchers are still studying the behavior of the alloy under steady state conditions to derive models to be applied for highly transient situations. Finally, under most forming conditions, the alloy behaves like a suspension of interacting solid particles so that this type of model can reasonably predict the filling of the mould and the optimum forming parameters. Development of two-phase models has been carried out more recently, but there is still much work to be done to introduce them in computer codes for the prediction of segregation during a forming operation. However, whatever be the type of model, pertinent microstructural parameters have to be brought in, which still requires some effort for their determination. On this aspect, X-ray microtomography is a powerful tool, which should be particularly appropriate for this determination despite the relative demands of the experiments.

Semisolid metal (SSM) processing is a hybrid technology combining features of both casting and forging that enables the production of near net-shape components of superior properties and surface finish. It was developed, from a discovery made at the Massachusetts Institute of Technology in the early 1970s, by Spencer et al. [1] that stirring of alloys during solidification led to a change in the solidifying microstructure resulting not only in a dramatic lowering of the apparent viscosity of the semisolid slurry, but also facilitating two-phase homogeneous flow at quite high fractions solid. More detail of this discovery and the effects of shear rate cooling rate and fraction solid can be found in Part II. The process of stirring alloys during solidification to produce non-dendritic solid within a slurry, and then injecting this slurry directly into a die as in liquid metal die casting, was called “rheocasting” by the MIT researchers and that name has largely stuck. Rheocasting started out as the preferred process route for industrial production and a new company formed byMIT and a group of industrial partners “Rheocast Corp.” designed and built several large-scale rheocasters for production of both aluminium and copper-base alloys. The first major customer of the technology and for the large-scale rheocasters produced by Rheocast Corp. was International Telephone and Telegraph (ITT) Corp.

Non-dendritic slurries were originally produced at MIT by mechanical stirring during solidification of the alloy. This started out as a batch process, but was quickly developed into a continuous slurry generator, or rheocaster, in which liquid alloy was passed through an annular channel while being simultaneously cooled and stirred to exit as a semisolid slurry. In the laboratory, these devices were scaled up to operate all the way up to steel processing temperatures and over the course of some 4 years; MIT researchers cast several thousand steel, aluminium and copper alloy parts using both the rheocasting and thixocasting process routes. Later, as commercial organisations became more involved, continuous rheocasters capable of producing hundreds of pounds per hour of aluminium and copper alloy slurry were developed and used to cast large quantities of copper alloy parts or aluminium extrusion billet. However, the complexity of operating multiple mechanical stirring units in commercial foundries and concerns about degradation of the equipment and contamination of the product quickly led to the development of electromagnetic (magneto hydrodynamic or MHD) stirring deep in the sump of a DC casting unit. The attraction of the MHD process was the potential to avoid gas pick up while providing vigorous stirring. This has now become the standard industrial route for the manufacture of SSM bar feedstock for use in thixocasting. The originalMHD process, developed and patented by Olin Corp. and assigned to ITT Inc., involved rotational flow about the vertical axis of a DC caster (Fig. 10.1).

Until recently, much of the SSM technology was highly proprietary and unknown to the general casting world. In the last several years, however, sufficient information has become available to make it possible to identify a number of alternative technical solutions that have been implemented around the world for the commercial production of high quality parts. Of particular note, it is now clear that real-time controlled machines are not an absolute pre-requisite for the production of safety critical or other high quality parts. Many parts are being produced in high volume and with excellent results on standard three-phase injection machines ((a) pre-set injection velocity, (b) pre-set ramp and (c) pre-set consolidation time under pressure before final ejection), as indeed were the very first semisolid die castings produced in the MIT laboratories. Rather, the key ingredients of a successful SSM system approach are: A reliable source of consistent quality raw material An appropriate delivery system for semisolid material to the casting machine A diaphragm or alternative approach to strip oxides from the slug’s surface or delivery system and eliminate or minimise the entrapment of oxides within the formed parts (see Figure 11.1) A powerful, repeatable, injection system capable of generating sufficient static pressure to feed solidification shrinkage throughout the freezing cycle Appropriate processes and controls to optimise the heat-treatability of the formed parts Coupled with intelligent die design to facilitate both turbulent-free filling and adequate feeding of shrinkage, these features have a proven record of producing high quality parts over long time periods.

Semisolid Metal Processes generate near net-shape end products with a high degree of precision because they use hard steel tooling and high injection pressures. This combination results in a high level of replication of the die contours and generally the NADCA standards for “precision tolerances” can be achieved on a regular basis. It is important whenever possible to eliminate sharp corners and provide proper draft and radii that assist the smooth and complete filling of the die cavity under speeds associated with high production cycles. Such provisions will tend to extend die life as well as allow the manufacturers to make use of the natural advantages of the SSM processes. Wall thicknesses typically range down as low as 1 mm, depending on the alloy system used, part size and shape and of course the intended application. Apart from the linear dimensional tolerances, other tolerances will be additive, such as “Parting line, Moving die components, Angularity, Concentricity, Parting line shift, Draft requirements, Flat requirements, Cored holes for cut threads, Cored holes for formed threads, Cored holes for pipe threads, and Machining stock allowance” tolerances. The NADCA publication no: 403 on “Product specification standards for die castings produced by the semisolid and squeeze casting processes” provides a full list of tolerances for a variety of alloys systems [NADCA].

As mentioned above, the premium cost of raw material has been an encumbrance to the widespread adoption of the technology. While niche producers, such as Vforge continue to grow their market share and specialised automotive products manufacturers, such as Madison Kipp Corp. continue to prosper, for semisolid metal processing to become once again a key player in the large-scale production of automotive components, a manageable solution to the cost of raw material must be demonstrated. Much is discussed currently within the field of various slurry-based approaches, but so far none are known to have made a serious inroad into automotive production where cost is as paramount as quality and performance. In the meantime, slurry technology shows promise as a means to lower the cost of conventional die castings by providing a partially solid charge material for parts already designed as die castings and thereby accelerating production and improving quality, but once again the jury remains out as to the trade-off between cost of the technology (both capital cost and operating cost) versus the realised benefits. In addition, although SSM of high melting point alloys offers exciting possibilities and tremendous potential, and has already been part of the original work of over 30 years ago [13], it is still at present in the research and development stage. More recent activities have produced complex shaped parts in steels as well as providing favourable direct economic comparisons with conventional forgings, as well as further development in ceramic or hybrid die materials [14]. These activities have now been consolidated under a European consortium with a brief to commercialise the thixoforming of high temperature alloys using a variety of steel grades as a starting point [15].