To pour or not to pour: the dilemma of assessing your aluminum melt's cleanliness.

How to determine a melt's inclusion content prior to pouring continues to trouble metalcasters. Here a status report on assessment methods.

As the saying goes, the more things change, the more they stay the same. And the same problem that foundrymen tinkering with aluminum had in the 1800s when the metal was developed still faces today's metalcasters. The problem...what to do to keep your metal "clean," and perhaps more important, determining when it isn't.

Several presentations at the 4th AFS International Conference on Molten Aluminum Processing, held November 12-14, 1995, in Orlando, examined assessing melt cleanliness. While process variables can be minimized, a key hurdle affecting many aluminum foundrymen is knowing "on-line" when the metal they are about to pour is free of contaminants, particularly inclusions.

According to I.R. Hughes, N-Tec, "Manufacturers will only remain competitive if they can detect potentially defective products (castings) early in the manufacturing process and act quickly to reduce the incidence of rejects to commercially acceptable levels."

"With the increasing use of scrap materials," said R.I.L. Guthrie, McGill Univ., "the inclusion concentration of the melt increases significantly. By use of 50-80% scrap, the inclusion concentration without a subsequent melt treatment can reach a high level (up to 35 ppm)."

"One of the most troublesome problems in producing aluminum castings is inclusion related defects," said J.R. Schmahl, Selee Corp. "The complex nature of the metalcasting process leads to many potential sources for inclusion formation, making the control of inclusions difficult to achieve."

While melt inclusions can be formed by a variety of factors, including salts, fluxes, grain refiners, eroded molding materials and refractories, this article examines those caused by oxides that contaminate clean aluminum.

Inclusions

Inclusions and oxide films in the bulk liquid (along with those formed by poor metal treatment) are most responsible for inconsistent casting production and batches of rejected castings, said Hughes.

The most prevalent source of inclusions, according to Schmahl, is through oxidation reactions with air. Molten aluminum reacts readily with oxygen in the air to form aluminum oxide ([Al.sub.2][O.sub.3]), which actually protects against further oxidation. Any disturbances to the protective oxide skin, however, result in additional aluminum oxide formation as the skin breaks. These oxides are typically of a skin-type morphology, although they may also be in small particle or flake form.

In addition to [Al.sub.2][O.sub.3], magnesium oxide (MgO) and magnesium aluminate spinel (Mg[Al.sub.2][O.sub.4]) are the most prevalent oxides. Figure 1 shows examples of these typical inclusions.

These broken films, according to Paul Crepeau, GM Powertrain Group, will likely show up in dross, an area deserving careful attention in the quest for clean aluminum. Dross is mainly composed of crumpled films that encapsulate a significant amount of unoxidized aluminum. Dross also contains inclusions scavenged from the liquid bath, flux constituents, ash, sludge particles and oxidation products from the contained metallic aluminum.

Wet dross, which is predominantly metallic aluminum alloy enveloped by oxide films, allows free communication between the liquid bath and oxides in the dross. Wet dross can easily bleed inclusions into molten aluminum because the surface energy barrier is already overcome.

Suspended inclusions create additional problems, Crepeau added, due to their effect on the precipitation of hydrogen. Despite their greater density, oxide inclusions easily stay in suspension. Inclusions, which tend to be small and/or irregularly shaped, can become crumpled and float in the bath like newspapers blown on a windy day.

Inclusions - even small ones - suspended in the aluminum bath interact with dissolved hydrogen and nucleate gas porosity during solidification. Guthrie added, "[Al.sub.2][O.sub.3], Mg[Al.sub.2][O.sub.4] and MgO inclusions generally have poor wettability with molten aluminum, and may provide potential heterogenous sites for pore nucleation. Certain local structural defects of oxide films and clusters, such as gaps, cracks and concave interfaces, will further facilitate heterogenous pore nucleation. Therefore, it can be expected that the porosity of a casting structure will increase proportionally as the total number of inclusions increases."

"As little as 5 ppm oxygen in the form of entrained aluminum films or other inclusions can significantly reduce properties by nucleating gas porosity during solidification or by acting as planar defects in the finished part," said Crepeau.

Effects of Dirty Metal

The results of inclusion-contaminated melt are many: poor machinability (hard spots), increased porosity, reduced corrosion resistance, loss of mechanical properties (especially in fatigue and ductility); a lack of pressure tightness, requiring weld repair or impregnation; surface finish unsuitable for anodizing and perhaps even for painting or varnish; and reduced fluidity. In other words, inclusions in aluminum can have a profound impact on casting scrap.

"Currently, metal is rarely checked for inclusion cleanliness during melting and holding," reported J.L. Roberge, Centre Technique des Industries de la Fonderie. "And yet alumina skins have dramatic effects on the mechanical properties of the castings."

The Need to Assess

"Since product consistency is often more important than maximum quality," said Hughes, "it is essential that the foundry engineer can quantitatively monitor molten metal cleanliness at each stage of the manufacturing process."

While the off-line metallographic examination of a solidified aluminum sample can provide an indication to the foundry's adherence to process control, foundrymen must be aware if its melt meets certain cleanliness standards before pouring.

At the 3rd International Conference on Molten Aluminum Processing held in 1992, Diran Apelian, Worcester Polytechnic Institute, reported: "What the industry desperately needs is an evaluation test on the factory floor, on an online basis, yielding information to the operator and allowing him to make adjustments to the molten metal stream prior to casting." At that time, he noted that the recently-developed Liquid Metal Cleanliness Analyzer (LiMCA) and the Porous Disc Filtration Apparatus (PoDFA, a pressure filtration technique) were both shop floor methods that held considerable promise for on-line metal cleanliness control.

In addition to the LiMCA and pressure filtration techniques, other tools have been developed as opportunities for metalcasters to better assess their aluminum melts.

LiMCA

As shown in Fig. 2, the LiMCA unit (marketed by Bomem, Inc.) is based on the electrical resistive pulse principle. Its probe consists of a heat-resistant sampling tube that draws molten aluminum under vacuum through a small orifice. A constant current is maintained through the orifice with a DC power supply and two electrodes, one placed inside the sampling tube, the other located in the surrounding melt. Any passage of a nonconducting particle (inclusion) through the orifice causes a drop in the electrical resistance, resulting in a voltage pulse.

A signal processing program developed at McGill Univ. calculates the particle density and size distribution by counting the number of pulses over time and by measuring their amplitudes. The resulting inclusion measurement is expressed as a volumetric concentration in ppm.

While Guthrie said it's a good qualitative test, particle sizes differ from melt to melt, and the detection limit of the LiMCA is limited due to the possible blocking of the tube with inclusions of greater sizes. "During LiMCA sampling, the orifice of the sample tube often blocks-particularly with the increased use of scrap material-probably by the collection of large oxides (films and clusters) around the orifice, resulting in lowered and inconsistent inclusion measurements."

The LiMCA delivers an immediate response, said Roberge, but it concerns small inclusions and seems best suited to measurement in continuous casting spouts. For example, in an experiment by McGill, an orifice size of 20 [[micro]meter] was chosen to avoid blockage experienced by smaller sizes. However, they said that the number of particles in a melt's total inclusion content too large to pass through this orifice may be as high as 20-60%, resulting in the LiMCA only measuring 40-80% of a melt's total inclusions.

Guthrie, who compared LiMCA to the pressure filtration technique, noted that both are useful tools for melt analysis and provide a wealth of detail about inclusion concentration, particle size distribution and type. But by the formation of a cake on the filter disc, "the pressure filtration sample is able to capture most of the inclusions suspended in the melt, while the LiMCA sample can measure only one part of the total inclusions."

Pressure Filtration Technique

One way to quantitatively measure the quality (metallurgical cleanliness) of aluminum alloys, said Hughes, is the pressure filtration method. It involves filtering a molten aluminum sample (of a known amount) under pressure through an ultrafine filter, resulting in a buildup of solid inclusions on the filter surface. By metallographically counting inclusions and measuring the area they occupy on a section of the filter cake, the inclusion level is expressed with a unit of inclusion area per kilogram of filtered melt ([mm.sup.2]/kg).

While this method is quantitative and useful, it isn't truly on-line, and is noted as time-consuming and expensive by Hughes. Guthrie added that a mixture of inclusions and other dispersed structural features on the solidified filter cake makes it difficult to accurately assess the total inclusion area. Furthermore, micro-porosity often occurs in the cross-section of the filter cake, particularly in dirty melts where inclusions and hydrogen are high. This leads to greater discrepancies during counting and measuring the inclusion area.

Prefil

Still under development at N-Tec, the Prefil technique builds on a quantitative understanding of cake filtration and how molten metal (containing different mixtures of inclusions), flows through an ultrafine filter [ILLUSTRATION FOR FIGURE 3 OMITTED]. But unlike traditional pressure filtration tests that require off-line diagnostic analysis, according to Hughes, a metal quality "footprint" can be defined for each foundry or product, which enables the foundryman to run within a preset molten metal cleanliness band.

Two years of pressure filtration tests to measure molten metal cleanliness have resulted in a database of the relationships between inclusion content and filtration rate behavior. Inclusions built-up on the filter surface were metallographically examined, identified and counted. The process concentrates the inclusions 5000-10,000 times, which results in total inclusion contents of 0.005 [mm.sup.2]/kg (very clean) and 50 [mm.sup.2]/kg (very dirty) being measured.

Quantitative metallography of pressure filtration residues is used in conjunction with existing in-house quality criteria to establish the relevant footprint. This correlation of inclusion contents with the shape of the filtration weight vs. time curves provides results that enable foundrymen to obtain immediate information on the quality of molten metal on a day-to-day basis without the need for continuous metallographic analyses.

The Prefil machine consists of a simple, two-stage pressure cell, a refractory crucible containing the filter and a digital balance to record filtrate weight against time. The specially designed crucible doesn't require preheating, enabling the test to be carried out under close temperature control with a high degree of repeatability. Measurements are expressed in [mm.sup.2]/kg.

An automatic filtration cell is operated as a simple "Go/No Go" foundry test. The database of foundry metal quality enables different processes to be benchmarked. Data can be downloaded to a PC and used for off-line diagnosis of a problem, or to re-footprint a process change. Dedicated software is being developed for use in conjunction with an automated shop floor unit.

Qualiflash

The Centre Technique des Industries de la Fonderie has developed what they call "a simplified new way of estimating the quantity of oxides, basically alumina skins, in an aluminum alloy bath."

As seen in Fig. 4, the Qualiflash determines melt cleanliness through the clogging of an extruded ceramic filter by oxides. Molten metal enters a temperature-controlled shell 788-806F (420-430C) with a filter at the bottom. The passing metal then flows into a nomograph (a 10-stepped ingot mold).

Once the metal has stopped flowing through the filter, the degree of cleanliness is quantified by the quantity of metal that fills the ingot mold. The dirtier a bath, the faster the clogging, and thus less metal in the mold. Prior calibration makes it possible to judge whether the bath is clean enough for the work to be done or whether the risk of scrap is high. "This makes it possible to decide whether the bath is ready to pour or must undergo another deoxidation treatment," Roberge said.

For example, with a pouring temperature of 1328F (720C) on a 356 Sr-modified melt, the oxide level measured by image analysis ranged from 0 to 50% and the mass of metal passing through the filter was 2.7 to 0.8 kg. In the ingot mold, as the oxide level goes from 0 to 50%, the number of steps filled or covered decreased from 8 to 2. "The device," said Roberge, "therefore is highly sensitive to the oxide level."

The simple test is complete in only 20 sec, and is read instantly by counting the number of steps covered in the ingot mold. (About 10 min is needed between tests to prepare a new filter). Advantages, said Roberge, are that the test can be used by relatively unskilled personnel, and the only consumable that must be replaced before each measurement is the filter.

K-Mold

The K-Mold device is a fracture test developed in Japan that has been used in the U.S. for fewer than 10 years. The advantages of using fracture tests for assessing melt cleanliness, according to D.E. Groteke, Q.C. Design, are low cost, quick results, low level of operator skill required, sampling flexibility and sample retention.

The K-Mold [ILLUSTRATION FOR FIGURE 5 OMITTED] produces a flat plate casting, with four notches cast into its cope surface. The notches serve as fracture points in the final shape, and the knife edges in the mold seem to improve the collection efficiency to capture the inclusion matter in the fracture zone through some "eddy" effect.

Multiple samples of the plates are poured in a preheated mold from the metal being evaluated. They are then fractured immediately by the technician. The preferred method of interpretation, said Groteke, is to consider each fracture face that contains one or more inclusions to be an "event." The opposite side of the fracture face will normally show evidence of the same inclusion, and is used to aid in interpreting the type of material, but isn't considered a separate "event." On that basis, there are four possible events per test.

The result (known as the K-value), is quantified by expressing the events as a ratio of total fractures examined. For example, if two dirty fractures are found in a total of 20 fractures (five bars with four fracture planes), the ratio is 2/20, and the K-value is 0.1.

Normally, allowable ratios are established for each product family, with those demanding a higher level of metal quality specifying a lower level of K-value. Typically, a thin-wall casting that has pressure tightness requirements may not exceed values of 0.05 (no more than one inclusion in 20 fractures).

According to Groteke, the K-Mold, which is marketed by Christy Refractories, is an easy to use, rapid means of evaluating melt quality. In its simplest application, it provides visual evidence of a need for timely corrective action to resolve problems with contaminated metal, and further provides a method for confirming the effectiveness of that action. Others, however, have reported difficulty with small round inclusions.

RELATED ARTICLE: Filtering: Added Security from Inclusions

The tools listed throughout this article can assist in your assessment of whether a melt is clean and help you make adjustments. One method to prevent inclusions from entering the mold cavity is the use of in-mold filters (placed in gating systems), as well as those placed directly in the furnace.

According to A. Matthews, Hamilton Porcelains, the three most common filters currently used in aluminum foundries are refractory cloth, steel screens and ceramic foam filters. Schmahl noted that in-mold filtration, which grew in acceptance in the late 1970s with the use of ceramic foam filters, separates inclusions by drawing the molten aluminum through a porous filter. Three modes of inclusion capture are generally used: screening, deep bed and cake filtration.

SEM/EDS analysis of spent ceramic foam filters indicates that both solid and liquid inclusion retention occurs, said Schmahl. He noted that filter pore size selection, sizing calculation and filter placement must be given careful consideration. It is critical to select the proper filter pore size to meet quality requirements, and to select the proper sized filter to prevent filter flow restriction and filter blockage.

Discussing an alternative material, Matthews said the ceramic cellular filter (used in iron foundries) provides low resistance to flow and consistent flow rates. Data also shows that it too is as effective as other materials in removing inclusions.

Studying the results of ceramic foam filters, Guthrie reported that the efficiencies of inclusion removal of a 30 ppi filter varies from 27% to 87%. "Although the average efficiency of inclusion removal by filters is quite reasonable," he reported, "the required final level of melt cleanliness for high quality casting products may not be obtained with such filters alone, because of the wide range of filtration performance."

Another filtration option, although not yet widespread, are continuous, reusable in-furnace filters to remove particulate inclusions before they reach the pouring ladle. The technology, said D. Neff, Metaullics Systems, LP, is based on the use of bonded particle filters, which can be reused because of their strength and durability, thermal conductivity and long-term corrosion resistance.

Bonded particle filters can be adapted as filter pumps in reverberatory melters; vertical gate filters in holding furnaces; box filters in holders, pots and crucibles; and stalk tube filters in low pressure permanent mold furnaces. The filtration efficiency results from the media's lower porosity, greater tortuosity and the scientific nature of the binder itself, which provides a "fly paper" effect to capture inclusions.

As used in reverberatory furnaces, a vertical gate filter separates the melting chamber from the tapout, dipout or pump well. They are usually installed by creating holding slots in a refractory lining or by adding guides inboard of the refractory wall to hold the filter in place. In most situations, the large surface areas afforded by the filter mean that when metal is taken out of the well, the metal flows or refills preferentially through the filter itself and not through the filter edges or the slots.

Depending on the application, these filters can last from several weeks to several months.