Preventing porosity in aluminum castings.

Porosity is caused by shrinkage and hydrogen, and this article reveals several keys for controlling this defect.

The first part of this series described how hydrogen enters castings during liquid solidification and its negative effects on casting quality. This month's article focuses on the other two ways porosity occurs in aluminum--shrinkage and, more typically, the combination of shrinkage and hydrogen.

Shrinkage

Shrinkage occurs during solidification as a result of volumetric differences between liquid and solid states. For foundrymen, the differences in liquid and solid volume are their greatest concern, while the temperature-dependent contraction after solidification is most important to die and patternmakers. For most aluminum alloys, volumetric shrinkage during solidification is about 6%.

To understand how shrinkage voids form, imagine a sphere without risers and picture how solidification takes place. Once the shell of the sphere solidifies and assumes enough strength to resist collapse, the continued process of cooling and solidification results in substantial tensile stresses in the liquid pool.

The solidified shell contracts at the rate dictated by the coefficient of thermal contraction for the solid, and the volume occupied by the liquid changes as a function of liquid contraction and solidification volume change, which additively are much greater. While the liquid struggles to maintain coherency, tensile forces ultimately exceed surface tension forces associated with the liquid-solid interface. A void will then form.

There are two variations in which the solidified shell lacks the integrity to resist these negative pressures. In the first case, the shell is coherent but weak, and localized collapse of the shell occurs to compensate for the volumetric change. This is an example of solid feeding. There may be no contained shrinkage void, only a surface depression. More typically, due to the imperfect fluid flow of the solid, the surface shrinkage depression can usually be associated with subsurface voids.

Elements that contribute to elevated temperature strength like iron, copper and nickel increase resistance to surface collapse of this kind. But since these elements also increase the solidification range, concentrated surface shrinkage forms.

This shrinkage causes the localized failure of the shell, allowing the gravimetric draining of interdendritic liquid into the liquid pool that remains. The interdendritic draining process continues during subsequent solidification as long as surface venting remains effective. This condition often is referred to as wormhole shrinkage.

Localized shell failure in commercial casting isn't random. It occurs at locations of heat and stress concentration. In the idealized example of the sphere, it would occur at the top because of gravity effects.

Contained shrinkage voids in our example occur when material solidification conditions permit the formation of a solidified shell of sufficient strength and thickness to resist the negative pressures increasing within the solidifying core.

The precipitation of hydrogen bubbles and the presence of oxides within the solidifying pool reduce the stresses at which shrinkage voids are initiated. This influences the location and distribution of the shrinkage voids that form. Oxides and other nonmetallics resist feeding through fluidity effects, and when present as large particles and films, these unwanted materials may block liquid metal flow during solidification.

The sphere example doesn't complete the big picture. It ignores the fact that, as soon as solidification begins, contact with the imaginary mold is lost in increasingly larger surface areas through circumferential contraction. It also is a poor example for examining the effects of localized hot spots created either by selective mold heating (flow path, impingement, design effects) or by liquid metal distribution, which typically contributes to shrinkage defects.

Displacing Shrinkage

If a chill is applied to the location of surface collapse or wormhole shrinkage as a sole corrective measure, defect formation simply moves to an adjacent area. Increasing chill size or moving the chill's location doesn't eliminate the defect--it just appears in another area that becomes the most susceptible to void formation.

Risers prevent shrinkage formation by maintaining a path for fluid flow from the higher heat mass and pressure of the riser to the encased liquid pool.

Shrinkage displacement by liquid takes place in four modes of feeding: liquid, mass, interdendritic and solid feeding. Of these, mass feeding and interdendritic feeding are of greatest importance.

Mass feeding is the displacement of liquid occurring in the absence of substantial resistance. In these cases, pressure at the solidification interface and pressure in the riser system are essentially equivalent. Pressure drop develops, however, as obstructions to the feeding path form. The progressive development of the dendrite network and localized solidification result in increased resistance to fluid flow until the pressure at the solidification front is reduced to zero, a conditional requirement for shrinkage void formation.

Shrinkage occurs in many forms. Distributed voids or microshrinkage are found between dendrite arms, which are formed as a result of failure during the last stages of interdendritic feeding. Centerline (or piping) voids result from gross directional effects--for example, when large liquid pools are isolated within the casting during solidification.

Studies to differentiate between mass and interdendritic feeding show mass feeding takes place over a much shorter thermal range than might be expected. Interdendritic feeding predominates over much of an alloy's solidification range so that controlled directional solidification becomes increasingly important as the solidus-liquidus range increases.

Solidification Range

Solidification range has always been correlated with castability and feedability. Ratings of these generalized characteristics by practical experience as well as from what is available in theoretical treatments confirm this direct, if not precise, correlation.

In alloys 319, 332 and 355, rapid cooling leads to the distribution of voids in the grain boundaries while slow cooling results in interdendritically distributed shrinkage. Shrinkage in 443 and 356 alloys, characterized by narrower solidification ranges, was found to be more localized at both solidification rates. In all cases, voids first began to form at temperatures corresponding to 65-75% solid.

For short solidification range alloys, such as 413 and 356, there is an improved opportunity for establishing soundness through directional solidification. Defects may more typically take the form of localized voids as opposed to distributed shrinkage porosity. When distributed porosity occurs, it is usually smaller in amount and pore size. These alloys may be characterized by a higher proportion of mass feeding relative to interdendritic feeding and are, therefore, less susceptible to the formation of shrinkage voids.

Solidification Distance

The tendency for interdendritic shrinkage is strongly affected by solidification distance (the dimension between solid and liquid zones). Achieving control over solidification fronts in such a way that solidification distance is uniformly reduced for this purpose is limited.

Solidification distance is inherently a function of solidification range. Wide solidification range alloys are more susceptible to shrinkage and porosity. Microporosity results from the higher proportion of feeding taking place interdendritically over a broader range of temperatures above the solidus. In the case of shrinkage, the influence of the relative multiplicity of feeding paths and feed path tortuosity (each of which increases with solidification distance) has generally been accepted.

Controlling Shrinkage

Shrinkage tendencies vary in proportion with the fourth root of the pressure. Unless extremely high pressures effective at the solidification interface are imposed, increasing pressure has little effect on shrinkage occurrence.

Riser effectiveness is derived more from temperature gradients than from the metallostatic head, although it cannot be ignored. Open risers are more effective than closed risers, and riser mass and position are largely thermal considerations.

Lower pouring temperatures in molds of finite heat extraction capacity establish more effective solidification gradients, and riser insulation promotes sustained gradients during solidification.

Improved modification and refinement of aluminum/silicon alloys, improved grain refinement and reduced oxide contents all improve feedability to reduce shrinkage tendencies.

Shrinkage Plus Hydrogen

For most "real-world" situations, independent distinctions between hydrogen and shrinkage porosity have little practical importance, since effects of one on the other are inevitable. When the complicating influence of oxide effects is added to hydrogen-shrinkage interactions, a third dimension of solidification defects technology comes into play.

For example, small amounts of dissolved hydrogen may significantly increase pore size compared to the same defects caused by shrinkage alone. The effects of hydrogen and shrinkage are interactive and may be disproportionately additive.

Hydrogen and shrinkage porosity are functions of melt characteristics and solidification conditions, and they affect defect morphology, distribution, pore size and pore volume fraction. Interaction is reduced by using chills or other techniques to correct either condition, but it cannot be totally eliminated.

The location where shrinkage porosity occurs depends on differential pressures arising from resistance to liquid metal flow on different dimensional scales.

By definition, shrinkage voids initiate in areas of low relative or zero pressure. Reduced pressure always results in a reduced threshold hydrogen value, leading to a facilitated precipitation of hydrogen into the void formed by shrinkage. In addition, solidification results in the solute enrichment of the liquid phase at the solidification front so that local hydrogen concentration and pressure are altered to promote hydrogen precipitation.

It generally is believed that hydrogen voids are rounded, smooth surface defects, while shrinkage voids invariably have the dendritic, jagged appearance that characterizes the dendrite structure. This conventional wisdom, however, is often wrong.

Hydrogen porosity can conform to dendrite arm regions, which gives bubble formation the characteristic appearance of a shrinkage void. Shrinkage occurring under extremely low gradients may assume a smooth-walled configuration. Likewise, the precipitation of hydrogen into a shrinkage void undergoing formation influences defect surface morphology. The interaction of hydrogen and shrinkage most certainly affects pore density, pore size and volume fraction of pores that form during solidification.

Adding Hydrogen

Intentionally adding hydrogen to counteract the more harmful effects of surface shrinkage on casting acceptability has become a common but often regrettable practice. For parts requiring only cosmetic as-cast appearance, it seems practical to add hydrogen by any number of means to improve superficial quality. But for parts requiring structural integrity, machining, leak resistance, or other specific mechanical or physical characteristic, the intentional addition of hydrogen is usually a mistake.

The precipitation of unusual amounts of hydrogen offsets the negative pressures that develop during solidification. The equalization of internal and external pressures brought about by hydrogen precipitation alone, or by hydrogen precipitation into internal shrinkage voids as they begin to form, minimizes the tendency for surface collapse and wormhole shrinkage formation. However, it also alters the size and distribution of voids in a manner that sacrifices internal integrity for improving external appearance.

Nucleation of hydrogen pores takes place as a function of oxide contamination. Oxide and other forms of nonmetallic contamination influence shrinkage pore formation. Because of fluid and thermal dynamics in mold filling and solidification, the basis for an accurate mathematical solidification analysis and prediction does not exist, although progress is moving closer to that objective.

Layered Feeding

Layered or sequential feeding in castings can be exploited to improve casting results. The first metal entering the cavity begins solidifying and is fed by the immediately adjacent molten metal layers entering the mold cavity--not by the riser system.

In sand castings, the last liquid to freeze is typically not localized along the centerline. When the gradient is low and the freezing range large, liquid-solid mushy zones may exist throughout the casting in various stages of solidification. Changes in fraction solid from surface to center may be small.

Nevertheless, localized gradients and the availability of thermally differentiated liquid at or near the solidification interface during the dynamic intervals comprised by the mold-filling process result in unexpected soundness in areas where shrinkage voids might otherwise be expected to occur.

This principle is routinely applied in the casting of wrought alloy ingot by continuous and discontinuous direct chill casting processes. In these casting operations, the solidifying interface is constantly fed by the newly introduced thermally differentiated alloy, and a degree of heat flow equilibrium is established to provide solidification conditions that assure minimum solidification zone growth. Solidification is always accompanied by unlimited liquid feed with adequate thermal gradient for the promotion of structural soundness.

Foundrymen can exploit these principles in gating designs that promote layering effects without resorting to the more costly methods that employ the same concepts in premium engineered casting work.

Controlling Porosity

Porosity in aluminum castings (whether hydrogen voids, shrinkage, or the more usual defects that can be associated with both conditions), can be understood and prevented by applying appropriate measures involving: melt treatment, gating/risering design, the effective use of variable heat extraction techniques, and the principles of directional and layered feeding.

Analytical tools permit the accurate definition of defects. Foundrymen have the ability to understand the nature of defect formation. With this knowledge, the occurrence of porosity in engineered castings can be prevented or controlled.

Foundrymen need not be confused by defect appearance or distribution. Nor should corrective action required for elimination or minimization of casting defects be complicated by inaccurate theories concerning defect formation.

Following these simple rules results in a more predictable casting quality:

* Hydrogen content must be controlled below threshold levels corresponding to solidification conditions.

* Shrinkage can be limited or eliminated by the application of the principles of directional solidification and directional feeding.

* The interactive effects of hydrogen and shrinkage must be considered in the analysis of defects, and in the development of appropriate corrective actions when unacceptable levels of porosity are experienced.

Rooy is a former manager of metallurgy and quality assurance at The Aluminum Company of America.

References

E.L. Rooy, E.F. Fischer, "Control of Aluminum Casting Quality for Vacuum Solidification Tests," AFS Transactions, vol 76, pp 237-240 (1968).

Q.T. Fang, P.N. Anyalebechi, "Effects of Solidification Conditions on Hydrogen Porosity Formation in Aluminum Alloy Castings," TMS Annual Meeting of Light Metals, p 477 (1988).

K.J. Brondyke, P.D. Hess, "Interpretation of Vacuum Gas Test Results for Aluminum Alloys," Transactions of AIME, vol 230, p 1542 (1964).